This is the accepted version of the journal article: Fernández-Orth, Dietmar; Miró, Elisenda; Brown-Jaque, Maryury; [et al.]. «Faecal phageome of healthy individuals : Presence of antibiotic resistance genes and variations caused by ciprofloxacin treatment». Journal of antimicro- bial chemotherapy, Vol. 74, Issue 4 (April 2019), p. 854-864. 11 pàg. DOI 10.1093/jac/dky540 This version is available at https://ddd.uab.cat/record/285165 under the terms of the license 1 Fecal phageome of healthy individuals; presence of antibiotic resistance genes and 2 variations caused by ciprofloxacin treatment 3 Dietmar FERNÁNDEZ-ORTH1, Elisenda MIRÓ2, Maryury BROWN-JAQUE3, Lorena 4 RODRÍGUEZ-RUBIO3, Paula ESPINAL2, Judith RODRIGUEZ-NAVARRO2,4, Juan 5 José GONZÁLEZ-LÓPEZ4,5, Maite MUNIESA3, Ferran NAVARRO2, 4* 6 1 7 ISGlobal, Hospital Clínic - Universitat de Barcelona, Barcelona, Spain 2 8 Servei de Microbiologia, Hospital de la Santa Creu i Sant Pau, Institut d'Investigació 9 Biomèdica Sant Pau, Sant Quintí 89, E-08041 Barcelona, Spain. 10 3Department of Genetics, Microbiology and Statistics, Universitat de Barcelona, 11 Diagonal 643, Annex, Floor 0, E-08028 Barcelona, Spain. 4 12 Departament de Genètica i de Microbiologia, Universitat Autònoma de Barcelona, 13 Barcelona, Spain. 5 14 Department of Clinical Microbiology, Hospital Vall d'Hebron, Vall d'Hebron Institut 15 de Recerca (VHIR), Pg Vall d’Hebron 119-129, 08035 Barcelona, Spain. 16 Corresponding author: Dr. Ferran Navarro. Phone: +34 93 5537297. Fax: 935537287. 17 e-mail: Fnavarror@santpau.cat 18 Running title- Phageome and antibiotics 19 Note: 20 - The authors Dietmar Fernández-Orth and Elisenda Miró contributed equally as first 21 authors 22 - The authors Maite Muniesa and Ferran Navarro contributed equally as senior authors 23 24 Synopsis 25 Objectives 26 Antimicrobial resistance genes (ARGs) can be transferred by means of mobile 27 genetic elements (MGEs), which play a critical role in the dissemination of resistance in 28 the bacterial community. ARG transmission within MGEs has been reported in 29 plasmids and transposons but less frequently in bacteriophages. Here, the bacteriophage 30 fraction of seven human fecal samples was purified and deep-sequenced to detect the 31 presence of ARGs in the phage particles. 32 Methods 33 Seven fecal samples (five from healthy individuals, and two from a patient 34 before and after receiving ciprofloxacin treatment) were used to extract phage DNA, 35 which was purified and then sequenced in a MiSeq (Illumina). Generated reads were 36 checked for quality and assembled, and the generated contigs analyzed with Kraken, 37 PHASTER, VirSorter and Prokka. Some genes were also validated by qPCR. 38 Results 39 Analysis of the purified phage DNA by Kraken identified from 4 to 266 viruses 40 in the samples. The viral fraction corresponded mainly to the order Caudovirales, 41 including phages from the Siphoviridae and Myoviridae families. Bacterial genes 42 associated with antimicrobial resistance were detected in the viral DNA, as confirmed 43 by qPCR. Higher densities of ARG-carrying phage particles were observed in the post- 44 versus pre-ciprofloxacin treatment sample. 45 Conclusions 46 The finding of ARGs in phage particles supports the description of phages as 47 mobile elements contributing to the dissemination of bacterial antibiotic resistance and 48 suggests ciprofloxacin treatment may play a role in the release of ARG-carrying 49 particles, thereby increasing resistance. 50 51 Introduction 52 Descriptions of the lower gastrointestinal tract microbiome have focused chiefly 53 on bacterial components1,2. However, recent research has suggested that the viral 54 fraction also contributes to shaping the microbiome structure3–6. 55 Within the virome, the phage fraction is the most abundant, with an active 56 phageome reported to comprise 35-2800 viruses,1,4,5 and many viral particles carry 57 bacterial DNA. 7–9 The great abundance of phages infecting all sorts of bacterial genera 58 suggests that phages and phage-derived particles are a mechanism of genetic exchange 59 that may play an important role in human health and disease. 6,10 60 Antibiotic treatment is associated with the rise of antimicrobial resistant bacteria 11 1 61 and affects microbiome composition and species abundance. As phages constitute 62 an important part of the microbiome, antibiotic perturbation might also affect phage 63 populations and promote their propagation rates, 12,13 thereby increasing phage-mediated 64 gene transfer. 65 Using a metagenomic approach, this study explored how phage particles may 66 actively contribute to antimicrobial resistance through the mobilization of antibiotic 67 resistant genes (ARGs) in human bacterial communities. When studying viromes, 14 68 contamination with bacterial DNA poses a serious drawback, therefore an optimized 69 protocol was used to ensure that DNA outside the phage capsids was completely 70 removed and that only packaged DNA was sequenced. 14 The role of phages in altering 71 ARG patterns in an individual receiving antibiotic treatment was also assessed. 72 73 Materials and methods 74 Fecal samples. Samples HC6, HC9, HC10, UB6 and UB25 were obtained from 75 five healthy individuals (aged 53, 39, 37, 27 and 26 years, respectively). The subjects 76 did not consume antibiotics in the three months prior to sample collection. Samples 77 PRECIP and POSTCIP were obtained from the same individual (aged 66 years): 78 PRECIP before taking antibiotics and POSTCIP four days after starting a 5-day 79 ciprofloxacin treatment (500 mg/12 hours) prescribed as prophylaxis before dental 80 surgery. The same individual supplied two independent fecal samples one year and one 81 year plus one week after the study, when no longer receiving any antibiotics. All 82 individuals live in Barcelona (Catalonia), none were involved in a food-borne outbreak 83 or showed any gastro-intestinal pathology or known infection, and none travelled 84 abroad in the three months before sampling except those providing samples HC9 and 85 HC10, who travelled to the Republic of Ecuador and Peru, respectively. 86 All samples were completely anonymized. The study was approved by the 87 Clinical Ethics Committee of the Hospital de la Santa Creu i Sant Pau (12/065/1350). 88 Informed consent was obtained for all individuals. 89 Purification of DNA from phage particles. Ten g of fecal samples were 90 homogenized 1:5 (w:v) in PBS by shaking for 30 minutes. 50 mL of the homogenate 91 was centrifuged at 3000 ×g and the supernatant was filtered through low protein- 92 binding, 0.22 µm pore-size membrane filters (Millex-GP, Millipore, Bedford, MA) to 93 remove cellular debris and bigger particles while allowing viral protein capsids, small 94 particles (including small vesicles) and free DNA to pass. The viral capsids in the 95 filtrates were 20-fold concentrated using protein concentrators (100 kDa Amicon Ultra 96 centrifugal filter units, Millipore, MA, US), at 3000 ×g following the manufacturer’s 97 instructions. Two mL concentrated suspensions were treated twice with chloroform 98 1:10, mixed by vigorous vortexing for 5 minutes and centrifuged at 16000 ×g for 10 99 minutes to disrupt bacterial membranes with the aim of breaking DNA-containing small 100 vesicles that might have passed through the 0.22 µm pore-size membrane filters. The 101 supernatant was treated with DNase (100 units/mL; Sigma-Aldrich, Spain) using the 102 reaction buffer provided by the manufacturer for 1 hour at 37ºC to eliminate non- 103 packaged DNA that might have passed the filters (bacterial DNA or DNA released from 104 vesicles after the chloroform treatment). The DNase was inactivated using the Stop 105 Solution provided by the manufacturer. 106 In this step, viral capsids remain intact and no vesicles should be observed, 107 whereas free DNA outside viral capsids should have been removed by the DNase 108 treatment. To confirm the absence of non-packaged bacterial DNA, an aliquot of the 109 phage suspension was taken at this stage and used as a template to amplify the selected 110 ARGs described below and eubacterial 16S rDNA by qPCR (Table 1). If the protocol 111 had successfully eliminated most of the vesicles as well as DNA outside the viral 112 capsids, both amplifications would be negative. 113 The protocols applied here for DNase treatment and DNase inactivation were 15 114 verified in our previous studies. Briefly, the DNase performance was verified by 115 adding known concentrations of an ARG that would not be detected by qPCR if 116 adequately eliminated by the DNase. DNase inactivation was previously confirmed by 117 adding known concentrations of an ARG immediately after the DNase inactivation step 118 and by amplifying this ARG by qPCR. An incomplete DNase inactivation would have 119 resulted in a reduced amplification of the ARG caused by DNase cleavage of the 120 primers in the qPCR assay, which was not observed. 15 121 122 Packaged DNA was extracted using a double extraction to achieve the required 123 purity. Firstly, the PowerSoil® DNA isolation kit (MO Bio. CA, US) was used, which 124 broke down the capsids and allowed the viral DNA to be recovered. Since the purity of 125 the extracted DNA was not yet optimal, according to the Bioanalyzer system (Agilent 126 Technologies, CA, US), further purification was carried out using the EZ1 Virus Mini 127 Kit v2.0 (Qiagen, CA, US) following the manufacturer’s instructions to a final volume 128 of 60 μL. 129 Sequencing. After purification, the DNA concentration was evaluated using a 130 Qubit® Fluorometer (Life Technologies. CA, US) and the DNA quality was further 131 confirmed using the 2100 Bioanalyzer system (Agilent Technologies, CA, US). 0.2 132 ng/μL of DNA was used to prepare the libraries. 133 Illumina libraries were generated following the Nextera XT (Illumina, Inc., San 134 Diego, CA, US) manufacturer’s protocol for paired-end libraries (2x150 bp). For 135 extension, 14 PCR cycles of 2.5 min were performed to increase the tagmentation 136 process. Libraries were purified and checked in a 2100 Bioanalyzer electrophoresis 137 instrument (Agilent Technologies, CA, US) and quantified in a Quantus™ Fluorometer 138 (Promega, WI, US). An equimolar pool of the seven samples was sequenced in an 139 Illumina MiSeq system with the Miseq Reagent Kit v2 (Illumina). 140 Bioinformatic analysis. Sequence reads were quality checked with FASTQC 141 v0.11.2 to detect any anomalies in the sequencing process. Reads were then de novo 16 142 assembled with default parameters using SPAdes v3.6.2. Contigs were classified 17 143 taxonomically with Kraken v0.10.5 using the standard Kraken database that 144 comprises NCBI taxonomic information, as well as the complete genomes in RefSeq for 145 the bacterial, archaeal, viral as well as phage domains. In parallel, contigs were 18 146 identified by comparison against the phage sequence database using PHASTER. 147 PHASTER classifies prophage sequences according to the percentage of the total 148 number of coding sequences (CDS) and the presence of specific phage-related words. 149 VirSorter was used to identify viral signals in the assembled contigs.19 150 Gene annotation was performed with Prokka 1.11,20 which establishes an e- 151 value of 10-6 within the different included databases. ARGs were searched with 152 ResFinder 2.1,21 which establishes a percentage of identity between CDS and the ARGs. 153 The minimum percentage of identical nucleotides shared between the best matching 154 ARG in the database and the corresponding sequence in a contig was set to 90 %. The 155 minimum percentage of overlapping nucleotides between a sequence and an ARG to 156 count as a hit for that gene was set to 60 %. 157 Sequences flanking the ARG detected by ResFinder were examined with 158 BLAST 22 to ascertain whether the sequences were phage or bacterial. 159 The search for 16S rDNA sequences in the contigs was performed with Metaxa2 23 160 software, which allows detection of the 16S DNA in metagenomic samples taking 161 into account the contigs generated in the novo assembling (done by Spades). 162 163 Standard PCR and qPCR procedures. PCRs were performed with an Applied 164 Biosystems 2720 Thermal Cycler (Applied Biosystems, Barcelona, Spain). ARGs from 165 the control strains were amplified by conventional PCR (Table 1). 166 Real-time qPCR assays using TaqMan hydrolysis probes for the different ARGs 24–27 167 (Table 1) were used under standard conditions as previously described. Those 168 ARGs analyzed by SYBR Green (Table 1) were quantified using PowerUp SYBR 169 Green Master Mix (ThermoFisher Scientifics, MA, US) following the manufacturer’s 170 instructions. 171 Availability of data and material 172 The data have been deposited with links to BioProject accession number 173 PRJNA392494 in the NCBI BioProject database 174 (http://www.ncbi.nlm.nih.gov/bioproject/392494). BioSample metadata are available in 175 the NCBI BioSample database (http://www.ncbi.nlm.nih.gov/biosample/) under 176 accession number SAMN07302947, SAMN07302946, SAMN07302945, 177 SAMN07302944, SAMN07302943, SAMN07302942 and SAMN07302940. 178 179 180 Results 181 Identification of DNA from phage particles 182 Study of DNA from phage particles gave a total of 11754439 reads (Table 2), 183 with an average depth of 1.68±0.18 Mb per sample and a sufficiently high quality for 184 our purposes. Trimming was not carried out, as the QC performed by FastQC showed 185 high quality reads, even in the last read positions. Sequence adaptors were also not 186 detected. The results obtained when trimming was performed were almost the same 187 (analysis not shown). Reads were first assembled with SPAdes and then taxonomically 188 classified with the Kraken Standard Database (Table 2). In addition, the assembled 18 189 contigs were used to perform a search for phage sequences with PHASTER (Table 2 190 and Supplementary File 1), which discriminates between sequences with a high 191 certainty of belonging to phages and those that are questionable or incomplete. 192 PHASTER identified a different number of phage-related contigs in the DNA from 193 phage particles compared to Kraken (Table 2). 194 20 195 The generated contigs were annotated using Prokka v1.11, which allowed 196 several phage genes to be identified (Table 2 and raw data in Supplementary File 2). 197 However, since genes in phage genomes can be shared among several phage taxa, 198 Prokka cannot determine if a given gene belongs to a specific phage. Data analysis 199 revealed phage structural genes (encoding capsid and tail), terminase and hydrolases 200 (Supplementary File 2). Also, Prokka detected ARG sequences in UB25, HC9, UB6, 201 PRECIP and POSTCIP (Supplementary File 2, blue). 202 Besides phage genes, Kraken analysis revealed numerous genes from different 203 bacterial species in the phage DNA fraction of the samples (Supplementary File 3). 204 Some genes were clearly of bacterial origin, while others, such as integrases or 205 ribonucleotide reductases, can be present in either phages or bacteria. 27,28 206 Contamination with non-packaged DNA was excluded by the controls, which 207 showed an absence of 16S rDNA and ARGs in the phage suspensions before 208 desencapsidation. Efficient DNA removal by DNase activity was confirmed and DNase 209 was completely inactivated by thermal treatment. 210 Additional information was obtained with VirSorter, which classified matches in 211 three categories. (i) Category 1 (“most confident” predictions) regions have significant 212 enrichment in viral-like genes and at least one hallmark viral gene detected. A hallmark 213 viral gene is typically of viral origin annotated as such (capsid gene, portal, terminase, 214 etc.). (ii) Category 2 regions correspond to “likely” predictions and are either enriched 215 in viral-like or non-Caudovirales genes, or have a viral hallmark gene detected. 216 Category 3 (“possible” prediction) was not considered. 217 218 Viral identification 219 According to the Kraken output, a high percentage of sequences in the viromes 220 (hereby understood as DNA extracted from the viral fraction, mostly phage particles) 221 belonged to bacterial or archaeal DNA (4.8%-29%), while the percentage of viral 222 sequences among the samples varied from 0.5% to 23.9% (Table 2 and Supplementary 223 File 3). The remaining sequences were not identified, or classed as uncharacterized 224 viruses (viral dark matter). 29 225 226 The distribution of the Kraken-identified viral sequences in the DNA from phage 227 particles revealed that the order Caudovirales, which includes most bacteriophages, was 228 the most abundant (Fig. 1, Table 3). Within this group, Siphoviridae was the most 229 prominent family, within which phages of lactic acid bacteria, including Skunalikevirus, 30 230 were the most abundantly identified by comparison against the databases. All the 231 sequences found in samples UB6 and UB25 belonged to the Siphoviridae. Sequences of 232 phages belonging to the Myoviridae family were only found in the PRECIP sample, 233 particularly those infecting Clostridium and Bacillus. The POSTCIP sample contained a 234 small percentage of Escherichia coli phages of the Podoviridae family. Based on 235 database sequences, HC6 showed a contig with 100 % identity with the phage ΦX174 236 from the family Microviridae, which infects E. coli, and HC9 contained Rhodococcus 237 phages. In addition, 50% of the viral sequences in HC9 belonged to the family 238 Mimiviridae, giant viruses that use Amoebae as hosts. 31 239 VirSorter also identified numerous phage genes belonging to Caudovirales in 240 several contigs of each sample (Table 2), particularly when categories 1 and 2 were 241 considered together. However, this is not indicative of the number of phages because it 242 is possible that several contigs belonged to the same phage. Data are available upon 243 request. 244 245 Detection of ARGs in the phage particles 246 The DNA from phage particles was searched for ARGs using ResFinder 2.1 and 247 Prokka 1.11 (Table 4). ResFinder identifies only a limited number of acquired ARGs, 248 while Prokka identifies any gene possibly involved in antibiotic or heavy metal 249 resistance, including those not exclusively involved in antibiotic resistance. The DNA 250 from phage particles was also searched for bacterial 16S rDNA using Metaxa2, and it 251 was detected in several contigs from all samples (Table 4). In contrast, 16S rDNA was 252 not detected in the samples before the capsids were broken down, as shown above. 253 254 Phage DNA in the samples was also analyzed by real-time qPCR targeting 255 specific ARGs (β-lactamases blaTEM, blaCTX-M-1 group and blaCTX-M-9 group, blaOXA-48, 256 methicillin resistance mecA, sulfonamide resistance sul1 and quinolone resistance qnrA 257 and qnrS). 24,25 These were selected for their clinical significance and because they have 24,25 258 previously been detected in human fecal viromes. We also included most of the 259 ARGs identified in the metagenomic analysis (tetracycline resistance tetW, β-lactamase 260 blaCFXA6, and macrolide resistance ermF and mefA). qPCR amplified at least one ARG 261 in all the samples except PRECIP (Table 4). Controls confirmed that the Taq 262 polymerase used was not contaminated with any ARG, as was previously reported for 32 263 Taq polymerase contaminated with blaTEM. The results obtained by metagenomic 264 analysis and qPCR did not always coincide. In general, metagenomics only detected the 265 most abundant ARGs (those showing a lower threshold cycle (Ct) by qPCR) (Table 4). 266 To shed light on the origin of the ARGs detected by ResFinder, the flanking regions 267 were analyzed by BLAST. The length of the flanking regions of each ARG varied 268 depending on the contig length, ranging from 127 to 5050 bp. Sequences showing more 269 than 95% identity with the fragment analyzed corresponded mostly to Bacteroides 270 fragilis or in a few cases other Bacteroides and bacterial species (Table 4). All 271 sequences corresponded with mobile genetic elements such as transposons or insertion 272 sequence elements, and specific phage sequences were not identified. 273 274 Comparison of samples from the same individual before (PRECIP) and 275 after (POSTCIP) ciprofloxacin treatment 276 PRECIP and POSTCIP samples showed differences in the metagenome structure 277 and viral content (Table 2). With a similar number of reads, the bacterial sequences 278 detected in POSTCIP did not change remarkably in terms of species. However, a greater 279 prevalence of genes from the Bacteroidetes group, lower prevalence of Actinobacteria 280 and higher viral prevalence and diversity was observed in the POSTCIP sample (Fig. 2). 281 All identified sequences in both samples corresponded to Caudovirales; PRECIP 282 presented a higher prevalence of Myoviridae, whereas in POSTCIP almost all were 283 Siphoviridae (Table 2; Fig. 2). 284 The PRECIP sample did not show any ARGs, and only Prokka identified two 285 efflux pump genes (involved in antimicrobial resistance) (Table 4). In contrast, ARGs 286 were detected in the POSTCIP sample by both metagenomic analysis and qPCR (Table 287 4). Notably, after the treatment with ciprofloxacin, POSTCIP showed low but detectable 288 levels of quinolone resistance genes (qnrA and qnrS). 289 To confirm that the higher ARG prevalence was caused by ciprofloxacin and to 290 gain insight into the background ARG levels in this individual, two additional samples 291 were taken one year and one year plus one week after the treatment. No ARGs were 292 detected by qPCR in phage DNA from these two samples. 293 294 Discussion 295 Studies of the phage community in human biomes have identified at least 1000 296 phage and prophage sequences, the majority belonging to double-stranded DNA viruses 297 of the Caudovirales order (Siphoviridae, Myoviridae, Podoviridae) and others to single- 298 stranded DNA viruses such as the Microviridae family. 33,34 An accurate analysis of the 299 DNA from phage particles greatly depends on the methods used, including the 300 extraction technique. In our study, we avoided concentration steps that could bias 301 species detection in the samples, 35 and phenol was not used, as it negatively affects the 302 purity of the extracted DNA. 25 303 Results also depend on the depth of sequencing. Illumina recommends starting 304 with a minimum of 100k reads, but here, as in similar studies, the number was increased 36 35 305 from 1 to 3 Mb to minimize the loss of sequences belonging to rare species. 306 Additionally, precise separation of the phage fraction is critical for avoiding chimeras, 307 which can be generated when analyzing total DNA in a metagenomic study. Shotgun 308 metagenomics thus provides a functional and detailed taxonomical classification of the 309 species composing the virome. 37 310 311 As no perfect tool exists for the detection of phage sequences or ARGs, it is 312 advisable to use a combination of different software, allowing contigs to be grouped 313 from different standpoints. We used SPAdes for the metagenomic assembly since it has 314 been favorably described compared with other options like Soap, Metavelvet, Omega or 315 Metahit.38 Different software tools will yield different results. For example, Kraken 316 includes any virus (not only phages, although these are the most abundant) and phage 317 sequence annotated as such in the databases. In contrast, VirSorter is more restrictive, 318 particularly for category 1 regions, and will only identify genes with a clear phage 319 origin. However, VirSorter allows the prediction of viral sequences outside the host 320 genome, even from short fragments. 19 321 322 One of the problems encountered when analyzing viromes is the possibility of 323 bacterial DNA contamination. The methods used were designed to prevent 324 contamination of the DNA extracted from phage particles. This contamination could 325 have three sources: bacterial or eukaryotic cells, free-cell DNA or DNA inside outer 39 326 membrane vesicles (OMV). Filtration eliminated bacterial or eukaryotic cells and 327 many OMV. The chloroform treatment, not always included in other virome studies, 328 was used to minimize the impact of DNA-containing OMV. 39 The residual presence of 329 vesicles containing DNA that passed through the filter and were not eliminated by the 330 chloroform treatment cannot be completely ruled out. However, their contribution to the 331 DNA pool analyzed should not be significant, based on previous studies using electron 24 332 microscopy in which the amount of phage particles visualized coincided with the 333 number of ARGs in the DNA extracted from these particles and no OMV were 334 observed. Finally, free DNA was removed using a strong DNase treatment, 15,40,41 335 demonstrated to efficiently eliminate non-packaged DNA. Absence of bacterial 336 DNA contamination was confirmed by the controls, which showed that free, non- 337 packaged DNA had been removed to the extent that 16S rDNA and ARGs were 338 undetectable by qPCR or metagenomics before the viral capsids were broken, whereas 339 these genes were later detected in the DNA from disrupted phage particles. 340 341 After rigorous efforts to eliminate all DNA outside the viral capsids, our results 342 still showed a wide fraction of bacterial DNA in phage particles, as previously observed 343 in viromes.35 Up to 70-80% of reads of viral metagenomes yield no significant matches, 3,42 344 possibly because few studies have focused specifically on phage sequencing, 345 resulting in a lack of specific phage databases and a low availability of phage sequences 43,44 346 in public databases in comparison with bacterial entries. In addition, there is a bias 347 toward the more studied human viruses, and a large fraction of virus sequences in the 29 348 databases correspond to uncharacterized viruses (viral dark matter) . Moreover, when 349 analyzing our sequences, we found that a number of prophage sequences within 350 bacterial genomes are erroneously annotated as bacterial, do not appear as phages when 351 searched in general databases, and cannot be identified in comparisons against the viral 352 database subsets. 353 354 Thus, assuming the absence of a significant proportion of bacterial DNA 355 contamination (but not completely ruled out), our hypothesis is that a high proportion of 356 bacterial DNA found within viral particles in this and other studies is the consequence 357 of generalized transduction or similar mechanisms able to package and spread 14,45–47 358 fragments of the bacterial genome in phage capsids. We recently showed that 359 phage particles harboring ARGs likely originate from bacterial strains in the 360 microbiome, and proposed that bacteria use this mechanism to spread their genome.24 361 This idea is supported by a recent study demonstrating that phage packaging in 362 Staphylococcus aureus frequently contributes to the mobilization of enormous 363 fragments of the genome of a single bacterial cell with great frequency. This new 364 mechanism, named “co-lateral transduction”, shows not only that phages can package 365 and mobilize bacterial genes, but that it is not a random process and that certain regions 366 of the S. aureus chromosome are preferentially hypermobile platforms of gene transfer. 47 367 368 If this mechanism also explains our observations, then using 16S rDNA as a 369 measure of contamination has its limitations, because like other bacterial genes, 16S 370 rDNA could also be packaged in the phage capsids and detected within the viral DNA. 371 Supporting our hypothesis, 16S rDNA was detected in the DNA from the phage 372 particles, and in more contigs than those showing ARGs. This was to be expected 373 considering that 16S rDNA should be more abundant in the bacterial population than 374 ARGs, perhaps with the exception of cells carrying ARGs located in multicopy 375 plasmids. Thus, monitoring 16S rDNA before breaking the capsids provides an accurate 376 way of verifying the absence of bacterial DNA contamination. 377 378 Special care should be taken when annotating ARGs. It has been repeatedly 379 demonstrated that a sequence showing homology with an ARG does not necessarily 380 indicate a true gene conferring antibiotic resistance.14,48 Therefore, the real extent of 381 ARGs in microbiomes and the associated risk of resistance transfer can be 382 overestimated. Nevertheless, the phageome is known to harbor ARGs,13,25,49 as shown 383 by the few studies on human phageomes 41,50 and environmental samples.51 384 The metagenomic data from the viral particles confirmed the presence of ARGs 385 in most samples. Parallel qPCR assays of the same DNA detected most of the ARGs 386 identified by metagenomic analysis, but the latter did not find all the ARGs detected by 387 qPCR. Although the Ct value of the qPCR is only indicative and varies among qPCR 388 assays, the output of the two methods largely coincided in the most abundant ARGs 389 (lower Ct), which suggests that high ARG densities are required for coverage by our 390 libraries. An exception was blaTEM, which was not detected by metagenomics despite a 391 Ct indicating higher abundance than other ARGs, perhaps because it was still close to 392 the coverage limit. 393 In one of the few studies measuring the efficiency of viral metagenomics in 394 comparison with PCR, the two approaches produced divergent results in different 395 matrices spiked with murine norovirus.35 When sequencing the virome, it is quite 52 396 difficult to estimate the level of coverage needed to detect less abundant genes. 397 Again, different software such as Prokka and ResFinder gave non-matching results. 398 Prokka combines the output of different tools that predict coding sequences, ribosomal 399 RNA, transfer RNA, signal leader peptides and non-coding RNA, whereas ResFinder 400 uses BLAST to identify ARGs. As each approach analyzes sequences from a different 401 standpoint, their results are not comparable, but both can provide valuable insights. 402 No specific phage sequences were found in the ARG flanking regions but 403 notably ARGs were located next to MGEs such as transposons or insertion sequences 404 (IS). This prevents the identification of the origin of the DNA (whether bacterial or 405 phage). As these elements are also present in the bacterial genomes, they could provide 406 platforms for ARG mobilization. The flanking regions were mostly identified as 407 belonging to Bacteroides fragilis. Since Bacteroidetes is the phylum most extensively 408 represented in our samples, this could explain why metagenomics detected more ARGs 409 related to these bacterial groups (Table 4). In contrast, β-lactamase genes (excluding 410 those from Bacteroidetes such as blaCFXA6) are mostly associated with 53 411 Enterobacteriaceae, present in the gut microbiome to a lesser extent than 412 Bacteroidetes. 54 Thus, β-lactamase genes or qnr genes belonging mostly to the less 413 abundant Enterobacteriaceae were only detected by the more sensitive qPCR assays and 414 not by metagenomics. 415 416 The samples analyzed before and after the ciprofloxacin treatment gave different 417 results. The PRECIP sample belongs to the 20% of fecal samples negative for ARGs in 25,55 418 phage DNA by qPCR described in previous work, whereas the POSTCIP sample 419 showed different mechanisms of resistance to various drugs. Although the absence of 420 ARGs in the normal state has not been explained, the presence of bacterial genes other 421 than ARGs (including 16S rDNA) in the viral DNA of the PRECIP sample indicates 422 that genomic DNA packaging occurred, although to a lesser extent than in POSTCIP. It 423 can be speculated that in this case some individual conditions repress the induction of 424 phage particles in general, or specifically those particles containing ARGs from the 425 microbiota, considering the microbiota as the main source of ARG-carrying phage 24 426 particles. Thus, ciprofloxacin might have reverted this repression or most probably 427 caused an increase in the induction of phage particles, which were detectable for a 428 period after the treatment. Absence of ARGs in phage DNA in samples taken one year 429 later suggests the microbiota of the individual had returned to the PRECIP values. 430 This observation supports the hypothesis that the presence of ARGs in the DNA 431 from phage capsids of POSTCIP was due to the antibiotic treatment. Stable 432 maintenance of ARGs and their elements in the absence of antibiotic selection has been 56 433 reported, but this is attributed to other genes in the same element conferring a 434 different kind of selective advantage. In contrast, some antibiotics are known to 435 exacerbate the transfer of their own resistance by the mobilization of genetic elements. 56 57 436 Higher rates of phage induction and recombination and greater abundance and 437 diversity have been reported in the presence of antibiotics in the swine gut microbiome. 58 438 Ciprofloxacin, one of these antibiotics, is reported to be an activator of the SOS 439 response and one of the most effective inducers of MGEs, including phages and derived 440 particles.12,59,60 441 In metagenomic studies, accurate methods and the use of a combination of 442 different software are strongly recommended to increase data output. Despite its 443 limitations, the metagenomic approach confirmed the presence of phage particles 444 containing several sequences of bacterial origin, including ARGs. The packaging 445 mechanisms are unknown but could involve transduction or other strategies by which 446 bacterial DNA is packaged, protected and spread. However, the association of ARGs 447 with phage particles points to the latter as a likely system of resistance dissemination 448 among the bacterial community. Future studies could elucidate the impact of phages on 449 the regulation of microbiota and the spread of multi-resistant bacteria as a problem of 450 global health. 451 452 Acknowledgements 453 Authors thank Prof. J. Jofre for his contribution to the study. Authors thank 454 ISGlobal, a member of the CERCA Programme, Generalitat de Catalunya. 455 456 Funding 457 The work was supported by Instituto de Salud Carlos III (ISCIII), co-financed by 458 the European Regional Development Fund (ERDF) “A Way to Achieve Europe”; 459 Spanish Network for Research in Infectious Diseases (REIPI RD12/0015/0017) ‘and’ 460 Spanish Ministry of Health (grant number FIS PI13/00329). PE is supported by ISCIII 461 (SB:CD15/00017). LRR is supported by the Beatriu de Pinós postdoctoral programme 462 of the Government of Catalonia's Secretariat for Universities and Research of the 463 Ministry of Economy and Knowledge. MBJ is supported with a grant from 464 COLCIENCIAS (Republic of Colombia). 465 466 Transparency of declaration 467 All authors have none to declare 468 469 References 470 1. Dethlefsen L, Huse SM, Sogin ML et al. The pervasive effects of an antibiotic 471 on the human gut microbiota, as as Revealed by Deep 16S rRNA Sequencing. PLoS 472 Biol 2008; 6: e280. 473 2. Gill SR, Pop M, Deboy RT et al. Metagenomic analysis of the human distal 474 gut microbiome. Science 2006; 312: 1355–9. 475 3. Džunková M, D’Auria G, Moya A. Direct sequencing of human gut virome 476 fractions obtained by flow cytometry. Front Microbiol 2015; 6: 955. 477 4. Pérez-Brocal V, García-López R, Nos P et al. Metagenomic analysis of 478 Crohn’s disease,patients identifies changes in the virome and microbiome related to 479 disease status and therapy, and detects potential interactions and biomarkers. Inflamm 480 Bowel Dis 2015; 21: 2515–32. 481 5. Norman JM, Handley S a, Parkes M et al. Disease-specific alterations in the 482 enteric virome in inflammatory bowel disease. Cell 2015; 160: 447–60. 483 6. Abeles SR, Pride DT. Molecular bases and role of viruses in the human 484 microbiome. J Mol Biol 2014; 426: 3892–906. 485 7. Navarro F, Muniesa M. Phages in the human body. Front Microbiol 2017; 8: 486 566. 487 8. Dutilh BE, Cassman N, McNair K et al. A highly abundant bacteriophage 488 discovered in the unknown sequences of human faecal metagenomes. Nat Commun 489 2014; 5: 37–43. 490 9. Minot S, Sinha R, Chen J et al. The human gut virome: inter-individual 491 variation and dynamic response to diet. Genome Res 2011; 21: 1616–25. 492 10. Dalmasso M, Hill C, Ross RP. Exploiting gut bacteriophages for human 493 health. Trends Microbiol 2014; 22: 399–405. 494 11. Ventola CL. The antibiotic resistance crisis: part 1: causes and threats. P T a 495 peer-reviewed J Formul Manag 2015; 40: 277–83. 496 12. Goerke C, Köller J, Wolz C. Ciprofloxacin and trimethoprim cause phage 497 induction and virulence modulation in Staphylococcus aureus. Antimicrob Agents 498 Chemother 2006; 50: 171–7. 499 13. Colomer-Lluch M, Jofre J, Muniesa M. Quinolone resistance genes (qnrA 500 and qnrS) in bacteriophage particles from wastewater samples and the effect of inducing 501 agents on packaged antibiotic resistance genes. J Antimicrob Chemother 2014; 69: 502 1265–74. 503 14. Enault F, Briet A, Bouteille L et al. Phages rarely encode antibiotic 504 resistance genes: a cautionary tale for virome analyses. ISME J 2017; 11: 237–47. 505 15. Colomer-Lluch M, Calero-Cáceres W, Jebri S, H et al. Antibiotic resistance 506 genes in bacterial and bacteriophage fractions of Tunisian and Spanish wastewaters as 507 markers to compare the antibiotic resistance patterns in each population. Environ Int 508 2014; 73: 167–75. 509 16. Bankevich A, Nurk S, Antipov D et al. SPAdes: a new genome assembly 510 algorithm and its applications to single-cell sequencing. J Comput Biol 2012; 19: 455– 511 77. 512 17. Wood DE, Salzberg SL. Kraken: ultrafast metagenomic sequence 513 classification using exact alignments. Genome Biol 2014; 15: 46. 514 18. Arndt D, Grant JR, Marcu A et al. PHASTER: a better, faster version of the 515 PHAST phage search tool. Nucleic Acids Res 2016; 44: 16–21. 516 19. Roux S, Enault F, Hurwitz BL et al. VirSorter: mining viral signal from 517 microbial genomic data. PeerJ 2015; 3: e985. 518 20. Seemann T. Prokka: rapid prokaryotic genome annotation. Bioinformatics 519 2014; 30: 2068–9. 520 21. Zankari E, Hasman H, Cosentino S et al. Identification of acquired 521 antimicrobial resistance genes. J Antimicrob Chemother 2012; 67: 2640–4. 522 22. Altschul SF, Gish W, Miller W et al. Basic local alignment search tool. J 523 Mol Biol 1990; 215: 403–10. 524 23. Bengtsson-Palme J, Hartmann M, Eriksson KM et al. METAXA2: improved 525 identification and taxonomic classification of small and large subunit rRNA in 526 metagenomic data. Mol Ecol Resour 2015; 15: 1403–14. 527 24. Brown-Jaque M, Calero-Cáceres W, Espinal P et al. Antibiotic resistance 528 genes in phage particles isolated from human feces and induced from clinical bacterial 529 isolates. Int J Antimicrob Agents 2018; 51: 434–42. 530 25. Quirós P, Colomer-Lluch M, Martínez-Castillo A et al. Antibiotic resistance 531 genes in the bacteriophage DNA fraction of human fecal samples. Antimicrob Agents 532 Chemother 2014; 58: 606–9. 533 26. Colomer-Lluch M, Imamovic L, Jofre J et al. Bacteriophages carrying 534 antibiotic resistance genes in fecal waste from cattle, pigs, and poultry. Antimicrob 535 Agents Chemother 2011; 55: 4908–11. 536 27. Lundin D, Gribaldo S, Torrents E et al. Ribonucleotide reduction - 537 horizontal transfer of a required function spans all three domains. BMC Evol Biol 2010; 538 10: 383. 539 28. Groth AC, Calos MP. Phage integrases: biology and applications. J Mol Biol 540 2004; 335: 667–78. 541 29. Thannesberger J, Hellinger H-J, Klymiuk I et al. Viruses comprise an 542 extensive pool of mobile genetic elements in eukaryote cell cultures and human clinical 543 samples. FASEB J 2017; 31: 1987–2000. 544 30. Rousseau GM, Moineau S. Evolution of Lactococcus lactis phages within a 545 cheese factory. Appl Environ Microbiol 2009; 75: 5336–44. 546 31. Aherfi S, Colson P, La Scola B et al. Giant viruses of amoebas: an update. 547 Front Microbiol 2016; 7: 349. 548 32. Song JS, Lee JH, Lee J-H et al. Removal of contaminating TEM-la beta- 549 lactamase gene from commercial Taq DNA polymerase. J Microbiol 2006; 44: 126–8. 550 33. Drulis-Kawa Z, Majkowska-Skrobek G, Maciejewska B et al. Learning from 551 bacteriophages - advantages and limitations of phage and phage-encoded protein 552 applications. Curr Protein Pept Sci 2012; 13: 699–722. 553 34. Łusiak-Szelachowska M, Weber-Dąbrowska B, Jończyk-Matysiak E et al. 554 Bacteriophages in the gastrointestinal tract and their implications. Gut Pathog 2017; 9: 555 44. 556 35. Hjelmsø MH, Hellmér M, Fernandez-Cassi X et al. Evaluation of methods 557 for the concentration and extraction of viruses from sewage in the context of 558 metagenomic sequencing. PLoS One 2017; 12: e0170199. 559 36. Cotten M, Oude Munnink B, Canuti M et al. Full genome virus detection in 560 fecal samples using sensitive nucleic acid preparation, deep sequencing, and a novel 561 iterative sequence classification algorithm. PLoS One 2014; 9. 562 37. Jovel J, Patterson J, Wang W et al. Characterization of the gut microbiome 563 using 16S or shotgun metagenomics. Front Microbiol 2016; 7: 459. 564 38. Vollmers J, Wiegand S, Kaster A-K. Comparing and evaluating metagenome 565 assembly tools from a microbiologist’s perspective - not only size matters! Rodriguez- 566 Valera F, ed. PLoS One 2017; 12: e0169662. 567 39. Biller SJ, Schubotz F, Roggensack SE et al. Bacterial vesicles in marine 568 ecosystems. Science (80- ) 2014; 343: 183–6. 569 40. Ross J, Topp E. Abundance of antibiotic resistance genes in bacteriophage 570 following soil fertilization with dairy manure or municipal biosolids, and evidence for 571 potential transduction. Appl Environ Microbiol 2015; 81: 7905–13. 572 41. Subirats J, Sànchez-Melsió A, Borrego CM et al. Metagenomic analysis 573 reveals that bacteriophages are reservoirs of antibiotic resistance genes. Int J Antimicrob 574 Agents 2016; 48: 163–7. 575 42. Ng TFF, Marine R, Wang C et al. High variety of known and new RNA and 576 DNA viruses of diverse origins in untreated sewage. J Virol 2012; 86: 12161–75. 577 43. Wagner J, Maksimovic J, Farries G et al. Bacteriophages in gut samples 578 from pediatric Crohn’s disease patients: metagenomic analysis using 454 579 pyrosequencing. Inflamm Bowel Dis 2013; 19: 1598–608. 580 44. Reyes A, Haynes M, Hanson N et al. Viruses in the faecal microbiota of 581 monozygotic twins and their mothers. Nature 2010; 466: 334–8. 582 45. Quirós P, Brown-Jaque M, Muniesa M. Spread of bacterial genomes in 583 packaged particles. Future Microbiol 2016; 11: 171–3. 584 46. Garcia-Aljaro C, Balleste E, Muniesa M. Beyond the canonical strategies of 585 horizontal gene transfer in prokaryotes. Curr Opin Microbiol 2017; 38: 95–105. 586 47. Chen J, Quiles-Puchalt N, Chiang YN et al. Genome hypermobility by 587 lateral transduction. Science 2018; 362: 207–12. 588 48. Martínez JL, Coque TM, Baquero F. What is a resistance gene? Ranking risk 589 in resistomes. Nat Rev Microbiol 2014; 13: 116–23. 590 49. Muniesa M, García A, Miró E et al. Bacteriophages and diffusion of beta- 591 lactamase genes. Emerg Infect Dis 2004; 10: 1134–7. 592 50. Modi SR, Lee HH, Spina CS et al. Antibiotic treatment expands the 593 resistance reservoir and ecological network of the phage metagenome. Nature 2013; 594 499: 219–22. 595 51. Balcazar JL. Bacteriophages as vehicles for antibiotic resistance genes in the 596 environment. PLoS Pathog 2014; 10: e1004219. 597 52. Sims D, Sudbery I, Ilott NE et al. Sequencing depth and coverage: key 598 considerations in genomic analyses. Nat Rev Genet 2014; 15: 121–32. 599 53. Coque TM, Baquero F, Canton R. Increasing prevalence of ESBL-producing 600 Enterobacteriaceae in Europe. Euro Surveill 2008; 13: 19044. 601 54. Eckburg PB, Bik EM, Bernstein CN et al. Diversity of the human intestinal 602 microbial flora. Science 2005; 308: 1635–8. 603 55. Larrañaga O, Brown-Jaque M, Quirós P et al. Phage particles harboring 604 antibiotic resistance genes in fresh-cut vegetables and agricultural soil. Environ Int 605 2018; 115: 133–41. 606 56. Salyers AA, Amábile-Cuevas CF. Why are antibiotic resistance genes so 607 resistant to elimination? Antimicrob Agents Chemother 1997; 41: 2321–5. 608 57. Johnson TA, Looft T, Severin AJ et al. The in-feed antibiotic carbadox 609 induces phage gene transcription in the swine gut microbiome. MBio 2017; 8. 610 58. Looft T, Johnson TA, Allen HK et al. In-feed antibiotic effects on the swine 611 intestinal microbiome. Proc Natl Acad Sci U S A 2012; 109: 1691–6. 612 59. Zhang X, McDaniel AD, Wolf LE et al. Quinolone antibiotics induce Shiga 613 toxin-encoding bacteriophages, toxin production, and death in mice. J Infect Dis 2000; 614 181: 664–70. 615 60. Ubeda C, Maiques E, Knecht E et al. Antibiotic-induced SOS response 616 promotes horizontal dissemination of pathogenicity island-encoded virulence factors in 617 staphylococci. Mol Microbiol 2005; 56: 836–44. 618 61. Lachmayr KL, Kerkhof LJ, Dirienzo AG et al. Quantifying nonspecific TEM 619 beta-lactamase (blaTEM) genes in a wastewater stream. Appl Environ Microbiol 2009; 620 75: 203–11. 621 62. Colomer-Lluch M, Jofre J, Muniesa M. Antibiotic resistance genes in the 622 bacteriophage DNA fraction of environmental samples. PLoS One 2011; 6. 623 63. Iwahara K, Kuriyama T, Shimura S et al. Detection of cfxA and cfxA2, the 624 beta-lactamase genes of Prevotella spp., in clinical samples from dentoalveolar 625 infection by real-time PCR. J Clin Microbiol 2006; 44: 172–6. 626 64. Chen J, Yu Z, Michel FC et al. Development and application of real-time 627 PCR assays for quantification of erm genes conferring resistance to macrolides- 628 lincosamides-streptogramin B in livestock manure and manure management systems. 629 Appl Environ Microbiol 2007; 73: 4407–16. 630 65. Volkmann H, Schwartz T, Bischoff P et al. Detection of clinically relevant 631 antibiotic-resistance genes in municipal wastewater using real-time PCR (TaqMan). J 632 Microbiol Methods 2004; 56: 277–86. 633 66. Sutcliffe J, Grebe T, Tait-Kamradt A et al. Detection of erythromycin- 634 resistant determinants by PCR. Antimicrob Agents Chemother 1996; 40: 2562–6. 635 67. Calero-Cáceres W, Melgarejo A, Colomer-Lluch M et al. Sludge as a 636 potential important source of antibiotic resistance genes in both the bacterial and 637 bacteriophage fractions. Environ Sci Technol 2014; 48: 7602–11. 638 68. Hunter N, Nadkarni MA, Jacques NA et al. Determination of bacterial load 639 by real-time PCR using a broad-range (universal) probe and primers set. Microbiology 640 2002; 148: 257–66. 641 642 Table 1. Oligonucleotides used in this study Amplimer LOQ of ARG Assay Oligonucleotide Sequence (bp) qPCR (Ct) Reference blaTEM qPCR-T UP CACTATTCTCAGAATGACTTGGT 85 30.5 61 LP TGCATAATTCTCTTACTGTCATG probe TEM FAM-CCAGTCACAGAAAAGCATCTTACGG-MGBNFQ blaCTX-M-1-group qPCR-T UP ACCAACGATATCGCGGTGAT 101 31.3 62 LP ACATCGCGACGGCTTTCT probe CTX-M-1 FAM–TCGTGCGCCGCTG- MGBNFQ blaCTX-M-9-group qPCR-T UP ACCAATGATATTGCGGTGAT 85 30.5 26 LP CTGCGTTCTGTTGCGGCT probe CTX-M-9 FAM – TCGTGCGCCGCTG- MGBNFQ blaOXA-48 qPCR-T UP CGGTAGCAAAGGAATGGCAA 133 32.4 24 LP TGGTTCGCCCGTTTAAGATT probe OXA-48 FAM – CGTAGTTGTGCTCTGGA- MGBNFQ blaCfxA6 qPCR-S UP GCGCAAATCCTCCTTTAACAA 802 31.5 63 LP ACCGCCACACCAATTTCG 64 ermF PCR-S UP CGACACAGCTTTGGTTGAAC 309 34.0 LP GGACCTACCTCATAGACAAG mecA qPCR-T UP CGCAACGTTCAATTTAATTTTGTTAA 92 33.8 65 LP TGGTCTTTCTGCATTCCTGGA probe mecA FAM-AATGACGCTATGATCCCAATCTAACTTCCACA-MGBNFQ mefA qPCR-S UP AGTATCATTAATCACTAGTGC 348 33.7 66 LP TTCTTCTGGTACTAAAAGTGG qnrA qPCR-T UP AGGATTGCAGTTTCATTGAAAGC 138 34.0 13 LP TGAACTCTATGCCAAAGCAGTTG probe qnrA FAM-TATGCCGATCTGCGCGA- MGBNFQ qnrS qPCR-T UP CGACGTGCTAACTTGCGTGA 118 34.6 13 LP GGCATTGTTGGAAACTTGCA probe qnrS FAM-AGTTCATTGAACAGGGTGA-MGBNFQ 67 sul1 qPCR-T UP CCGTTGGCCTTCCTGTAAAG 67 34.5 LP TTGCCGATCGCGTGAAGT probe sul1 FAM-CGAGCCTTGCGGCGG-MGBNFQ tetW qPCR-T UP GACGGACACCATGTTTTTGGA 62 33.7 This study LP AGGAAGTGACTGCCGCTTGA probe tetW AGCGTGGGATTACCA 16S rDNA qPCR-T UP TCCTACGGGAGGCAGCAGT 467 32.2 68 LP GGACTACCAGGGTATCTAATCCTGTT probe 16S FAM-CGTATTACCGCGGCTGCTGGCAC-MGBNFQ 643 qPCR-T: quantitative real time PCR using TaqMan hydrolysis probes. 644 qPCR-S: quantitative real time PCR using a SybrGreen assay. 645 UP: upper primer, LP: lower primer, FAM: 6-carboxyflouorescein reporter, MGBNFQ: Minor groove binding non-fluorescent quencher. 646 LOQ. Limit of quantification determined with the standard curve used in this study for each qPCR assay. 647 648 Table 2. Bioinformatic analysis of DNA from phage particles. Sample # Reads #Contigs Kraken PHASTER Prokka VirSorter Nº taxonsa Nº virusesb Nº of Nº of CDS Category 1c Category 2d phage Nº phage Nº phage Nº phage Nº phage (%) (%) phages genes contigse genesf contigse genesf PRECIP 1653277 11100 532 (4.8) 11 (2.0) 17 40 7162 7 159 51 1379 POSTCIP 1612020 18469 1689 (9.1) 182 (10.8) 12 41 10769 4 28 61 1014 HC6 1758529 6015 842 (14.0) 4 (0.5) 12 25 2297 2 7 13 391 HC9 1799296 4604 440 (9.6) 4 (0.9) 10 39 4050 1 21 31 968 HC10 1372547 3838 1113 (29.0) 266 (23.9) 7 31 2415 0 0 34 739 UB6 1954802 17814 3035 (17.0) 56 (1.8) 12 35 10419 6 79 48 1298 UB25 1603968 4949 781 (15.8) 24 (3.0) 8 21 6457 4 69 28 1022 a 649 All microorganisms: archaea, bacteria and viruses. b 650 viruses including phages. 651 c “most confident prediction”: high probability of phage identification. 652 d “likely”: good probability of phage identification. e 653 Nº of contigs where genes of phages have been identified. f 654 Nº of genes of phages identified in all contigs. 655 CDS: Coding regions. 656 657 Table 3. DNA from phage particles of the samples. Percentages of sequences identified by Kraken as different phages. Phages Hosts PRECIP POSTCIP HC6 HC9a HC10a UB6 UB25 Caudovirales 91 100 75 50 99.2 100 100 Siphoviridae 63.7 99.5 75 25 99.2 100 100 Skunalikevirus Lactococcus 27 31 50 31 21 29 C2likevirus Lactococcus 9 Unclassified Lactococcus 27 57 25 55.4 41 66 Lactobacillus 9 5 Leuconostoc 2 33 Enterococcus 2 Propionibacterium 2 Rhodococcus 25 Others 0.7 10 25 3.8 1 Myoviridae 27.3 Clostridium 9 Bacillus 9 Others 9.3 Podoviridae 0.5 Microviridae Enterobacteria 25 Mimiviridae Unclassified 50 Unassigned 9 0.8 a 658 subject HC9 travelled to the Republic of Ecuador; subject HC10 travelled to Peru 659 Table 4. ARGs, ARG flanking regions and 16S rDNA in the DNA from phage particles ARG ARG Flanking regions 16S rDNA ResFinde (length of the flanking fragment) # contigs qPCR(Ct) Prokka b r (GenBank Accession #) PRECIP --- MdtH, MdtK --- --- 7 POSTCIP mecA (32.76), qnrA (32.09), qnrS Aluminiumc, Bicyclomycin, CzcAc, DrrA, blaCfxA5 Bacteroides fragilis insertion 16 (32.94), tet(W) (25.35) Glyoxalase/Bleomycin, operon repressor, sequence IS614B (93 bp) MdtA, MdtB, MdtC, MdtE, MdtK, MexA, (AY769934) MexB, NorM, MarA, YheH, EmrK, SugE, TerB, TetO, TelA, Trypsin-resistant surface T6 protein, VanW tet(W) Faecalibacterium prausnitzii plasmid mobilization relaxosome protein MobC (5050 bp) (CP026548.1) HC6 blaTEM (28.45), blaCTX-M-1 group --- erm(F) No flanking regions, contig aligns 3 (32.09) a, blaCfxA6 (26.74), erm(F) totally (30.17), qnrA (34.29) a, qnrS (34.59) a, tet(W) (29.99) HC9 blaTEM (28.26), blaCTX-M-1 group CzcAc, TerB --- --- 6 (32.97)a, blaCTX-M-9 group (32.74) a, blaCfxA6 (28.88) a, HC10 blaCTX-M-1 group (32.86) qnrS --- --- --- 4 (30.12), tet(W) (31.94) UB6 blaTEM (28.43), blaCTX-M-9 group BmrA, CnrBc, CzcAc, DrrA, aadE No flanking region, contig aligns 13 (33.41) a, blaCfxA6 (22.43), mef(A) Fosmidomycin, Lincosamide, MarA, totally (30.15), sul1 (32.79), MecR1, MdtE, MdtK, TetM tet(W)(27.09) blaCfxA B. fragilis transposon Tn4555 TnpA, integrase, TnpC, excisionase, mobilization protein and unknown genes (1118 bp) (U75371.3) blaCfxA6 No flanking regions, contig aligns totally mef(A) B. fragilis strain HMW 615 transposon CTnHyb (304 bp) (KJ816753.1) tet(Q) HMW615 transposon CTnHyb (1836 bp)(KJ816753.1) tet(X) B. fragilis transposon Tn4351 (1836 bp)(M37699.1) UB25 blaTEM (28.21), blaOXA-48 (34.10)a, Fosmidomycin, CnrBc CzcAc, blaCfxA Bacteroides ovatus insertion 16 blaCTX-M-9 group (29.81), blaCfxA6 Lincosamide, MdtE, MdtK, TetM sequence IS4-like (127 bp) (27.73), mef(A) (26.21), qnrS (HE999703.1) (30.85), qnrA(34.25) a , sul1 (32.58), tet(W) (24.37) mef(A) B. fragilis transposon CTnHyb, (277 bp) (KJ816753.1) tet(Q) B. fragilis transposon Tn4351 (390 bp)(M37699.1) tet(X) B. fragilis transposon Tn4351 (390 bp)(M37699.1) 660 a Beyond the limit of quantification of the standard curve of each qPCR assay but still detectable. 661 b The annotation of specific proteins responsible for antibiotic resistance or the elements to which resistance is conferred are indicated. 662 c Proteins conferring resistance to metals. 663 Ct: Threshold cycle. 664 ARGs in bold indicate coincidence between qPCR and ResFinder results. 665 Figure legends 666 667 Figure 1. Pie Graphs of Kraken data for samples HC6, HC9, PRE-CIP and POSTCIP from 668 healthy individuals. Distribution and relative abundance of different bacteriophages identified in the 669 viromes with regard to the total viral sequences identified. Interactive pie graphs available upon 670 request. This figure appears in colour in the online version of JAC and in black and white in the 671 printed version of JAC. 672 673 Figure 2. Pie Graphs of Kraken data showing metagenomic comparison of the virome of the 674 two samples from the same individual taken before (PRECIP) and after (POSTCIP) ciprofloxacin 675 treatment. This figure appears in colour in the online version of JAC and in black and white in the 676 printed version of JAC.