Bacteria are able to pass each other genes that make them invulnerable to antibiotics. This exchange of genetic material, also called horizontal gene transfer, can turn otherwise harmless bacteria into drug-resistant ‘superbugs’. This is particularly problematic in hospitals, where bacteria use horizontal gene transfer to become resistant to several antibiotics and disinfectants at once, leading to serious infections that are difficult to treat.

How can scientists stop bacteria from sharing genes with one another? To answer this question, first it is important to understand how horizontal gene transfer happens in the bacteria that cause infections in hospitals. To this end, Evans et al. examined the genomes of over 2000 different bacteria, collected from a hospital over 18 months, for signs of horizontal transfer. First the experiments identified the genetic material that had potentially been transferred between bacteria, also known as ‘mobile genetic elements’. Next, Evans et al. examined the data of patients who had been infected with the bacteria carrying these mobile genetic elements to see whether horizontal transfer might have happened in the hospital.

By combining genomics with patient data, it was determined that many of the mobile genetic elements identified were likely being shared among hospital bacteria. One of the mobile genetic elements identified was able to provide resistance to several drugs, and appeared to have been horizontally transferred between bacteria infecting two separate patients.

The findings of Evans et al. show that the horizontal transfer of mobile genetic elements in hospital settings is likely frequent, but complex and difficult to study with current methods. The results of this study show how these events can now be tracked and analyzed, which may lead to new strategies for controlling the spread of antibiotic resistance.

Horizontal gene transfer (HGT) is a driving force behind the multidrug-resistance and heightened virulence of healthcare-associated bacterial infections (Lerminiaux and Cameron, 2019). Genes conferring antibiotic resistance, heightened virulence, and environmental persistence are often encoded on mobile genetic elements (MGEs), which can be readily shared between bacterial pathogens via HGT (Juhas, 2015). While rates of HGT are not well quantified in clinical settings, prior studies have shown that MGEs can mediate and/or exacerbate nosocomial outbreaks (Bosch et al., 2017; Jamrozy et al., 2017; Martin et al., 2017; Sheppard et al., 2016). Recent studies have also demonstrated that multidrug-resistant healthcare-associated bacteria share MGEs across large phylogenetic distances (Cerqueira et al., 2017; Hazen et al., 2018; Kwong et al., 2018). Understanding the dynamics of MGE transfer in clinical settings can uncover important epidemiologic links that are not currently identified by traditional infection control methodologies (Lerminiaux and Cameron, 2019; Schmithausen et al., 2019; Stadler et al., 2018).

Methods to identify and track the movement of MGEs among bacterial populations on short timescales are limited. Bacterial whole-genome sequencing has transformed infectious disease epidemiology over the last decade (Ladner et al., 2019), providing powerful new tools to identify and intervene against outbreaks (Sundermann et al., 2019b). Despite these advances, efforts to track MGE movement have focused almost exclusively on drug resistance and virulence genes (Cerqueira et al., 2017; Hardiman et al., 2016; Martin et al., 2017; Stadler et al., 2018), often ignoring the broader genomic context of the mobile elements themselves. Many studies rely on the identification of plasmid replicons, transposases, and other ‘marker genes’ (Orlek et al., 2017), an approach that oversimplifies the diversity of MGEs and may lead to incomplete or erroneous conclusions about their epidemiology. While querying databases containing curated MGE-associated sequences is useful for the rapid screening of clinical isolates for known MGEs, it will not capture novel MGEs. Additionally, whole-genome sequencing using short-read technologies generates genome assemblies that usually do not resolve MGE sequences, due to the abundance of repetitive elements that MGEs often contain (Arredondo-Alonso et al., 2017). Advances in long-read sequencing can mitigate this problem; hybrid assembly of short- and long-read sequence data allows the genomic context of chromosomal and extrachromosomal MGEs to be precisely visualized (Cerqueira et al., 2017; Conlan et al., 2014; George et al., 2017). Finally, studying the epidemiology of MGEs in clinical settings requires detailed individual-level patient clinical data, without which HGT occurrence in the hospital cannot be identified (Conlan et al., 2014).

Here, we performed an alignment-based screen for shared nucleotide sequences in a large and diverse collection of bacterial genomes sampled from infections within a single hospital over an 18-month time period. With this approach, we identified shared sequences that occurred in the genomes of bacteria belonging to different genera. Because they were identical, we suspect that these sequences recently transferred between bacteria within the hospital setting. Further analysis using long-read sequencing and reference-based resolution of distinct MGEs enabled us to precisely characterize MGE architecture and cargo, and to track MGE occurrence over time. Cross-referencing our results with available patient metadata allowed us to follow these elements as they emerged and were maintained among nosocomial bacterial populations.

Through this study, we have produced a high-resolution view of shared sequence and MGE dynamics among clinical bacterial isolates collected over an 18-month period from a single hospital. We identified, clustered, and characterized identical sequences found in multiple distinct genera, and in the process uncovered both expected and unexpected cases of shared sequence occurrence. We confirmed that some of the most common shared sequences identified were fragments of larger MGEs. We performed long-read sequencing to resolve these larger elements, and in doing so we characterized a large diversity of drug resistance-encoding plasmids. When we traced the presence of various plasmids over time, we found some that were likely transferred vertically along with the bacteria carrying them, and others that appeared to be transferred horizontally between unrelated bacteria.

Our study adds to the body of knowledge of HGT in hospital settings in new and important ways. We analyzed a large set of clinical isolates collected from a single health system, and used a systematic approach to identify shared nucleotide sequences, regardless of their type or gene content. While prior studies have used genomic epidemiology to study how HGT contributes to the transmission, persistence, and virulence of bacterial pathogens (Bosch et al., 2017; Martin et al., 2017; Schweizer et al., 2019; Valenzuela et al., 2007), the technical challenges of resolving MGEs from whole-genome sequencing data have limited the scope of these findings (Arredondo-Alonso et al., 2017). Furthermore, while rates of HGT between pathogenic bacteria have been quantified in vitro, very little information is currently available to assess rates of HGT in vivo or in clinical settings (Leclerc et al., 2019). Other studies have deliberately tracked HGT in healthcare settings by focusing either on mobile genes of interest, such as those encoding drug resistance (Cerqueira et al., 2017; Hardiman et al., 2016; Hazen et al., 2018), or on specific classes of MGEs (Savinova et al., 2019). Both of these approaches likely generate incomplete accounts of the extent of HGT in clinical settings. For this reason, we selected a pairwise alignment-based approach, whereby we only looked for identical sequences in the genomes of very distantly related bacteria. In doing so, we did not limit ourselves to only focusing on ‘known’ MGEs, and thus obtained a more accurate and comprehensive overview of the dynamics of HGT between bacterial genera in our hospital.

What might cause horizontally-transferred nucleotide sequences to be found at very high identity within phylogenetically distinct bacteria? Among many possible causes, we could consider the following: (1) the sequences we identified could have been recently transferred and not have had time to diverge from one another; (2) they could already be well adapted to optimally perform their functions; or (3) they could represent genetic elements that are highly intolerant to mutation. We suspect that our dataset contains all three cases. First, in the instances of likely within-patient HGT, both plasmids isolated from the same patient were nearly identical to one another. This suggests that if mutation rates of plasmids are similar to bacterial chromosomes, these plasmids would have transferred shortly before the bacteria were isolated. In both cases of likely within-patient HGT we also observed similar plasmids in the genomes of isolates from other patients, but we identified a likely route of transfer between patients only in the case where the subsequent plasmid was also nearly identical. This finding supports our theory that high plasmid identity is evidence of recent transfer. Second, the plasmids that we identified only in ST258 K. pneumoniae or in ST131 E. coli are likely well adapted to these lineages, perhaps because plasmid-imposed fitness costs have already been resolved through compensatory adaptations (San Millan, 2018). Third, the Tn7 transposon sequence we uncovered, which was identical in bacterial isolates from three different genera, was also identical to over two dozen publicly available genome sequences queried through a standard NCBI BLAST search. The insertion of the Tn7 transposon downstream of glmS in all of our isolates suggests TnsD-mediated transposition (Parks and Peters, 2009, p. 7), but the reason why the entire transposon sequence remains so highly conserved remains unclear.

The vast majority of shared sequences identified through our approach contained signatures of mobile elements, and our follow-up work demonstrated that these sequences could very likely move independently and assemble in a mosaic fashion on larger mobile elements like plasmids, integrative conjugative elements, and other genomic islands. Antibiotic resistance genes were present in only a subset of the shared sequence clusters we identified, which was somewhat surprising given how many resistance genes are known to be MGE-associated. Our follow-up analysis showed, however, that resistance genes were indeed highly prevalent among many of the MGEs that we resolved. This finding is consistent with a recent study of clinical K. pneumoniae genomes, which showed that while antibiotic resistance genes were largely maintained at the population level, they were variably present on different MGEs that fluctuated in their prevalence over time (Ellington et al., 2019). Finally, we were surprised by the large number of metal-interacting genes and operons within the shared sequences that we identified. While metal-interacting genes and operons have been hypothesized to confer disinfectant tolerance and increased virulence (Chandrangsu et al., 2017; McDonnell and Russell, 1999), precisely how these elements might increase bacterial survival in the hospital environment and/or contribute to infection requires further study.

Identification of risk factors and common exposures for HGT has previously been proposed (Conlan et al., 2014; Hardiman et al., 2016; Lerminiaux and Cameron, 2019; Pecora et al., 2015), but the results of prior efforts have been limited because large genomic datasets from single health systems with corresponding epidemiologic data have not been widely available (Struelens, 1998). The use of routine whole-genome sequencing for outbreak surveillance in our hospital has allowed us to begin to study how the horizontal transfer of MGEs might be similar or different from bacterial transmission. In addition to finding evidence of vertical transfer of plasmids accompanying bacterial transmission, we also identified several cases in which the same MGE was identified in two or more isolates of different sequence types, species, or genera. In some cases, these isolates were collected within days or weeks of one another. This finding highlights the frequent movement of MGEs between bacterial populations, particularly in hospitalized patients (Huddleston, 2014; Lerminiaux and Cameron, 2019), and points to the importance of pairing genome sequencing with epidemiologic data to uncover routes of MGE transmission.

There are several limitations to our study. First, the dataset that we used only contained genomes of isolates from clinical infections from a pre-selected list of species, and did not include environmental samples or isolates from patient colonization. In the case of between-patient plasmid transfer that we identified, we do not know exactly how the plasmid was transferred from Patient A to Patient B because we did not collect these intermediaries. Second, our method to screen for shared sequences based on cross-genus alignment was based on arbitrary alignment length and identity cutoffs. As expected, we detected more clusters at shorter alignment cut-offs, and we suspect that decreasing the identity threshold would also result in the identification of more and bigger clusters. Additionally, we did not consider sequences found in different bacteria within a single genus for the purposes of cluster identification. The cross-genus parameter we employed may have also artificially enriched the number of MGEs identified among Enterobacteriaceae, which are known to readily undergo HGT with one another (Cerqueira et al., 2017). Third, we assigned MGE presence relative to single reference sequences, and based our analysis on reference sequence coverage; subsequent MGEs that either gained additional sequence or rearranged their contents would still be assigned the same MGE, even though they may have diverged substantially from the reference MGE (Sheppard et al., 2016). Finally, this study was based exclusively on comparative genome analyses, and the MGEs we resolved from clinical isolate genomes were not tested for their capacity to undergo HGT in vitro.

In conclusion, we have shown how bacterial whole genome sequence data, which is increasingly being generated in clinical settings, can be leveraged to study the dynamics of HGT between drug-resistant bacterial pathogens within a single hospital. Our future work will include further characterization of the shared sequences and MGEs we resolved, assessment of sequence sharing across closer genetic distances (such as within-genus transfer), exploration of MGE and host co-evolution, and incorporation of additional epidemiologic information to identify shared exposures and possible routes for MGE transfer independent from bacterial transmission. Ultimately, we aim to develop this analysis into a reliable method that can generate actionable information and enhance traditional approaches to prevent and control multidrug-resistant bacterial infections.

Bacterial genome sequencing data have been deposited to relevant NCBI databases (SRA/GenBank).

1. 1. van Tyne D

   (2020) NCBI BioProject

   ID PRJNA609916. Systematic analysis of cross-genus horizontal gene transfer among bacterial pathogens in a single hospital.

   https://www.ncbi.nlm.nih.gov/bioproject/?term=PRJNA609916

1. 1. Altschul SF
   2. Gish W
   3. Miller W
   4. Myers EW
   5. Lipman DJ

   (1990) Basic local alignment search tool

   Journal of Molecular Biology 215:403–410.

   https://doi.org/10.1016/S0022-2836(05)80360-2

   • PubMed
   • Google Scholar
2. 1. Arredondo-Alonso S
   2. Willems RJ
   3. van Schaik W
   4. Schürch AC

   (2017) On the (im)possibility of reconstructing plasmids from whole-genome short-read sequencing data

   Microbial Genomics 3:e000128.

   https://doi.org/10.1099/mgen.0.000128

   • Google Scholar
3. 1. Bankevich A
   2. Nurk S
   3. Antipov D
   4. Gurevich AA
   5. Dvorkin M
   6. Kulikov AS
   7. Lesin VM
   8. Nikolenko SI
   9. Pham S
   10. Prjibelski AD
   11. Pyshkin AV
   12. Sirotkin AV
   13. Vyahhi N
   14. Tesler G
   15. Alekseyev MA
   16. Pevzner PA

   (2012) SPAdes: a new genome assembly algorithm and its applications to single-cell sequencing

   Journal of Computational Biology 19:455–477.

   https://doi.org/10.1089/cmb.2012.0021

   • PubMed
   • Google Scholar
4. 1. Bosch T
   2. Lutgens SPM
   3. Hermans MHA
   4. Wever PC
   5. Schneeberger PM
   6. Renders NHM
   7. Leenders A
   8. Kluytmans J
   9. Schoffelen A
   10. Notermans D
   11. Witteveen S
   12. Bathoorn E
   13. Schouls LM

   (2017) Outbreak of NDM-1-Producing Klebsiella pneumoniae in a dutch hospital, with interspecies transfer of the resistance plasmid and unexpected occurrence in unrelated health care centers

   Journal of Clinical Microbiology 55:2380–2390.

   https://doi.org/10.1128/JCM.00535-17

   • PubMed
   • Google Scholar
5. 1. Carattoli A
   2. Zankari E
   3. García-Fernández A
   4. Voldby Larsen M
   5. Lund O
   6. Villa L
   7. Møller Aarestrup F
   8. Hasman H

   (2014) In silico Detection and Typing of Plasmids using PlasmidFinder and Plasmid Multilocus Sequence Typing

   Antimicrobial Agents and Chemotherapy 58:3895–3903.

   https://doi.org/10.1128/AAC.02412-14

   • Google Scholar
6. 1. Cerqueira GC
   2. Earl AM
   3. Ernst CM
   4. Grad YH
   5. Dekker JP
   6. Feldgarden M
   7. Chapman SB
   8. Reis-Cunha JL
   9. Shea TP
   10. Young S
   11. Zeng Q
   12. Delaney ML
   13. Kim D
   14. Peterson EM
   15. O'Brien TF
   16. Ferraro MJ
   17. Hooper DC
   18. Huang SS
   19. Kirby JE
   20. Onderdonk AB
   21. Birren BW
   22. Hung DT
   23. Cosimi LA
   24. Wortman JR
   25. Murphy CI
   26. Hanage WP

   (2017) Multi-institute analysis of carbapenem resistance reveals remarkable diversity, unexplained mechanisms, and limited clonal outbreaks

   PNAS 114:1135–1140.

   https://doi.org/10.1073/pnas.1616248114

   • PubMed
   • Google Scholar
7. 1. Chandrangsu P
   2. Rensing C
   3. Helmann JD

   (2017) Metal homeostasis and resistance in Bacteria

   Nature Reviews Microbiology 15:338–350.

   https://doi.org/10.1038/nrmicro.2017.15

   • PubMed
   • Google Scholar
8. 1. Conlan S
   2. Thomas PJ
   3. Deming C
   4. Park M
   5. Lau AF
   6. Dekker JP
   7. Snitkin ES
   8. Clark TA
   9. Luong K
   10. Song Y
   11. Tsai YC
   12. Boitano M
   13. Dayal J
   14. Brooks SY
   15. Schmidt B
   16. Young AC
   17. Thomas JW
   18. Bouffard GG
   19. Blakesley RW
   20. Mullikin JC
   21. Korlach J
   22. Henderson DK
   23. Frank KM
   24. Palmore TN
   25. Segre JA
   26. NISC Comparative Sequencing Program

   (2014) Single-molecule sequencing to track plasmid diversity of hospital-associated carbapenemase-producing Enterobacteriaceae

   Science Translational Medicine 6:254ra126.

   https://doi.org/10.1126/scitranslmed.3009845

   • PubMed
   • Google Scholar
9. 1. Ellington MJ
   2. Heinz E
   3. Wailan AM
   4. Dorman MJ
   5. de Goffau M
   6. Cain AK
   7. Henson SP
   8. Gleadall N
   9. Boinett CJ
   10. Dougan G
   11. Brown NM
   12. Woodford N
   13. Parkhill J
   14. Török ME
   15. Peacock SJ
   16. Thomson NR

   (2019) Contrasting patterns of longitudinal population dynamics and antimicrobial resistance mechanisms in two priority bacterial pathogens over 7 years in a single center

   Genome Biology 20:184.

   https://doi.org/10.1186/s13059-019-1785-1

   • PubMed
   • Google Scholar
10. 1. Eyre DW
    2. Cule ML
    3. Wilson DJ
    4. Griffiths D
    5. Vaughan A
    6. O'Connor L
    7. Ip CLC
    8. Golubchik T
    9. Batty EM
    10. Finney JM
    11. Wyllie DH
    12. Didelot X
    13. Piazza P
    14. Bowden R
    15. Dingle KE
    16. Harding RM
    17. Crook DW
    18. Wilcox MH
    19. Peto TEA
    20. Walker AS

    (2013) Diverse sources of C. difficile infection identified on whole-genome sequencing

    The New England Journal of Medicine 369:1195–1205.

    https://doi.org/10.1056/NEJMoa1216064

    • PubMed
    • Google Scholar
11. 1. Garcillán-Barcia MP
    2. Redondo-Salvo S
    3. Vielva L
    4. de la Cruz F

    (2020) MOBscan: automated annotation of MOB relaxases

    Methods Mol Biol Clifton NJ 2075:295–308.

    https://doi.org/10.1007/978-1-4939-9877-7_21

    • Google Scholar
12. 1. George S
    2. Pankhurst L
    3. Hubbard A
    4. Votintseva A
    5. Stoesser N
    6. Sheppard AE
    7. Mathers A
    8. Norris R
    9. Navickaite I
    10. Eaton C
    11. Iqbal Z
    12. Crook DW
    13. Phan HTT

    (2017) Resolving plasmid structures in Enterobacteriaceae using the MinION nanopore sequencer: assessment of MinION and MinION/Illumina hybrid data assembly approaches

    Microbial Genomics 3:e000118.

    https://doi.org/10.1099/mgen.0.000118

    • PubMed
    • Google Scholar
13. 1. Gurevich A
    2. Saveliev V
    3. Vyahhi N
    4. Tesler G

    (2013) QUAST: quality assessment tool for genome assemblies

    Bioinformatics 29:1072–1075.

    https://doi.org/10.1093/bioinformatics/btt086

    • PubMed
    • Google Scholar
14. 1. Hardiman CA
    2. Weingarten RA
    3. Conlan S
    4. Khil P
    5. Dekker JP
    6. Mathers AJ
    7. Sheppard AE
    8. Segre JA
    9. Frank KM

    (2016) Horizontal transfer of Carbapenemase-Encoding plasmids and comparison with hospital epidemiology data

    Antimicrobial Agents and Chemotherapy 60:4910–4919.

    https://doi.org/10.1128/AAC.00014-16

    • Google Scholar
15. 1. Hazen TH
    2. Mettus R
    3. McElheny CL
    4. Bowler SL
    5. Nagaraj S
    6. Doi Y
    7. Rasko DA

    (2018) Diversity among bla_KPC-containing plasmids in Escherichia coli and other bacterial species isolated from the same patients

    Scientific Reports 8:10291.

    https://doi.org/10.1038/s41598-018-28085-7

    • PubMed
    • Google Scholar
16. 1. Huddleston JR

    (2014) Horizontal gene transfer in the human gastrointestinal tract: potential spread of antibiotic resistance genes

    Infection and Drug Resistance 7:167–176.

    https://doi.org/10.2147/IDR.S48820

    • PubMed
    • Google Scholar
17. 1. Jamrozy D
    2. Coll F
    3. Mather AE
    4. Harris SR
    5. Harrison EM
    6. MacGowan A
    7. Karas A
    8. Elston T
    9. Estée Török M
    10. Parkhill J
    11. Peacock SJ

    (2017) Evolution of mobile genetic element composition in an epidemic methicillin-resistant Staphylococcus aureus: temporal changes correlated with frequent loss and gain events

    BMC Genomics 18:684.

    https://doi.org/10.1186/s12864-017-4065-z

    • PubMed
    • Google Scholar
18. 1. Jia B
    2. Raphenya AR
    3. Alcock B
    4. Waglechner N
    5. Guo P
    6. Tsang KK
    7. Lago BA
    8. Dave BM
    9. Pereira S
    10. Sharma AN
    11. Doshi S
    12. Courtot M
    13. Lo R
    14. Williams LE
    15. Frye JG
    16. Elsayegh T
    17. Sardar D
    18. Westman EL
    19. Pawlowski AC
    20. Johnson TA
    21. Brinkman FS
    22. Wright GD
    23. McArthur AG

    (2017) CARD 2017: expansion and model-centric curation of the comprehensive antibiotic resistance database

    Nucleic Acids Research 45:D566–D573.

    https://doi.org/10.1093/nar/gkw1004

    • PubMed
    • Google Scholar
19. 1. Jolley KA
    2. Bray JE
    3. Maiden MCJ

    (2018) Open-access bacterial population genomics: bigsdb software, the PubMLST.org website and their applications

    Wellcome Open Research 3:124.

    https://doi.org/10.12688/wellcomeopenres.14826.1

    • PubMed
    • Google Scholar
20. 1. Jolley KA
    2. Maiden MC

    (2010) BIGSdb: scalable analysis of bacterial genome variation at the population level

    BMC Bioinformatics 11:595.

    https://doi.org/10.1186/1471-2105-11-595

    • PubMed
    • Google Scholar
21. 1. Juhas M

    (2015) Horizontal gene transfer in human pathogens

    Critical Reviews in Microbiology 41:101–108.

    https://doi.org/10.3109/1040841X.2013.804031

    • Google Scholar
22. 1. Komano T
    2. Fujitani S
    3. Funayama N
    4. Kanno A
    5. Sakuma K

    (1990) Physical and genetic analyses of IncI2 plasmid R721: evidence for the presence of shufflon

    Plasmid 23:248–251.

    https://doi.org/10.1016/0147-619X(90)90057-J

    • PubMed
    • Google Scholar
23. 1. Krzywinski M
    2. Schein J
    3. Birol I
    4. Connors J
    5. Gascoyne R
    6. Horsman D
    7. Jones SJ
    8. Marra MA

    (2009) Circos: an information aesthetic for comparative genomics

    Genome Research 19:1639–1645.

    https://doi.org/10.1101/gr.092759.109

    • PubMed
    • Google Scholar
24. 1. Kwong JC
    2. Lane CR
    3. Romanes F
    4. Gonçalves da Silva A
    5. Easton M
    6. Cronin K
    7. Waters MJ
    8. Tomita T
    9. Stevens K
    10. Schultz MB
    11. Baines SL
    12. Sherry NL
    13. Carter GP
    14. Mu A
    15. Sait M
    16. Ballard SA
    17. Seemann T
    18. Stinear TP
    19. Howden BP

    (2018) Translating genomics into practice for real-time surveillance and response to carbapenemase-producing Enterobacteriaceae: evidence from a complex multi-institutional KPC outbreak

    PeerJ 6:e4210.

    https://doi.org/10.7717/peerj.4210

    • PubMed
    • Google Scholar
25. 1. Ladner JT
    2. Grubaugh ND
    3. Pybus OG
    4. Andersen KG

    (2019) Precision epidemiology for infectious disease control

    Nature Medicine 25:206–211.

    https://doi.org/10.1038/s41591-019-0345-2

    • PubMed
    • Google Scholar
26. 1. Leclerc QJ
    2. Lindsay JA
    3. Knight GM

    (2019) Mathematical modelling to study the horizontal transfer of antimicrobial resistance genes in Bacteria: current state of the field and recommendations

    Journal of the Royal Society Interface 16:20190260.

    https://doi.org/10.1098/rsif.2019.0260

    • Google Scholar
27. 1. Lerminiaux NA
    2. Cameron ADS

    (2019) Horizontal transfer of antibiotic resistance genes in clinical environments

    Canadian Journal of Microbiology 65:34–44.

    https://doi.org/10.1139/cjm-2018-0275

    • PubMed
    • Google Scholar
28. 1. Manges AR
    2. Geum HM
    3. Guo A
    4. Edens TJ
    5. Fibke CD
    6. Pitout JDD

    (2019) Global extraintestinal pathogenic Escherichia coli (ExPEC) Lineages

    Clinical Microbiology Reviews 32:00135-18.

    https://doi.org/10.1128/CMR.00135-18

    • Google Scholar
29. 1. Marçais G
    2. Delcher AL
    3. Phillippy AM
    4. Coston R
    5. Salzberg SL
    6. Zimin A

    (2018) MUMmer4: a fast and versatile genome alignment system

    PLOS Computational Biology 14:e1005944.

    https://doi.org/10.1371/journal.pcbi.1005944

    • PubMed
    • Google Scholar
30. 1. Marsh JW
    2. Mustapha MM
    3. Griffith MP
    4. Evans DR
    5. Ezeonwuka C
    6. Pasculle AW
    7. Shutt KA
    8. Sundermann A
    9. Ayres AM
    10. Shields RK
    11. Babiker A
    12. Cooper VS
    13. Van Tyne D
    14. Harrison LH

    (2019a) Evolution of Outbreak-Causing Carbapenem-Resistant Klebsiella pneumoniae ST258 at a Tertiary Care Hospital over 8 Years

    mBio 10:01945-19.

    https://doi.org/10.1128/mBio.01945-19

    • Google Scholar
31. 1. Marsh JW
    2. Pacey MP
    3. Ezeonwuka C
    4. Ohm SL
    5. Snyder D
    6. Cooper VS
    7. Harrison LH
    8. Doi Y
    9. Mustapha MM

    (2019b) Clostridioides difficile: a potential source of NpmA in the clinical environment

    Journal of Antimicrobial Chemotherapy 74:521–523.

    https://doi.org/10.1093/jac/dky420

    • PubMed
    • Google Scholar
32. 1. Martin J
    2. Phan HTT
    3. Findlay J
    4. Stoesser N
    5. Pankhurst L
    6. Navickaite I
    7. De Maio N
    8. Eyre DW
    9. Toogood G
    10. Orsi NM
    11. Kirby A
    12. Young N
    13. Turton JF
    14. Hill RLR
    15. Hopkins KL
    16. Woodford N
    17. Peto TEA
    18. Walker AS
    19. Crook DW
    20. Wilcox MH

    (2017) Covert dissemination of carbapenemase-producing Klebsiella pneumoniae (KPC) in a successfully controlled outbreak: long- and short-read whole-genome sequencing demonstrate multiple genetic modes of transmission

    Journal of Antimicrobial Chemotherapy 72:3025–3034.

    https://doi.org/10.1093/jac/dkx264

    • PubMed
    • Google Scholar
33. 1. McDonnell G
    2. Russell AD

    (1999) Antiseptics and disinfectants: activity, action, and resistance

    Clinical Microbiology Reviews 12:147–179.

    https://doi.org/10.1128/CMR.12.1.147

    • PubMed
    • Google Scholar
34. 1. Orlek A
    2. Stoesser N
    3. Anjum MF
    4. Doumith M
    5. Ellington MJ
    6. Peto T
    7. Crook D
    8. Woodford N
    9. Walker AS
    10. Phan H
    11. Sheppard AE

    (2017) Plasmid classification in an era of Whole-Genome sequencing: application in studies of antibiotic resistance epidemiology

    Frontiers in Microbiology 8:182.

    https://doi.org/10.3389/fmicb.2017.00182

    • PubMed
    • Google Scholar
35. 1. Parks DH
    2. Chuvochina M
    3. Waite DW
    4. Rinke C
    5. Skarshewski A
    6. Chaumeil PA
    7. Hugenholtz P

    (2018) A standardized bacterial taxonomy based on genome phylogeny substantially revises the tree of life

    Nature Biotechnology 36:996–1004.

    https://doi.org/10.1038/nbt.4229

    • PubMed
    • Google Scholar
36. 1. Parks AR
    2. Peters JE

    (2009) Tn7 elements: engendering diversity from chromosomes to episomes

    Plasmid 61:1–14.

    https://doi.org/10.1016/j.plasmid.2008.09.008

    • PubMed
    • Google Scholar
37. 1. Pecora ND
    2. Li N
    3. Allard M
    4. Li C
    5. Albano E
    6. Delaney M
    7. Dubois A
    8. Onderdonk AB
    9. Bry L

    (2015) Genomically informed surveillance for Carbapenem-Resistant Enterobacteriaceae in a health care system

    mBio 6:e01030.

    https://doi.org/10.1128/mBio.01030-15

    • PubMed
    • Google Scholar
38. 1. Pruitt KD
    2. Tatusova T
    3. Maglott DR

    (2007) NCBI reference sequences (RefSeq): a curated non-redundant sequence database of genomes, transcripts and proteins

    Nucleic Acids Research 35:D61–D65.

    https://doi.org/10.1093/nar/gkl842

    • PubMed
    • Google Scholar
39. 1. Quan H
    2. Sundararajan V
    3. Halfon P
    4. Fong A
    5. Burnand B
    6. Luthi JC
    7. Saunders LD
    8. Beck CA
    9. Feasby TE
    10. Ghali WA

    (2005) Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data

    Medical Care 43:1130–1139.

    https://doi.org/10.1097/01.mlr.0000182534.19832.83

    • PubMed
    • Google Scholar
40. 1. San Millan A

    (2018) Evolution of Plasmid-Mediated antibiotic resistance in the clinical context

    Trends in Microbiology 26:978–985.

    https://doi.org/10.1016/j.tim.2018.06.007

    • PubMed
    • Google Scholar
41. 1. Savinova TA
    2. Bocharova YA
    3. Lazareva AV
    4. Chebotar IV
    5. Mayanskiy NA

    (2019) Integron diversity in blavim-2-carrying carbapenem-resistant clinical Pseudomonas aeruginosa isolates

    Russian Clinical Laboratory Diagnostics 64:497–502.

    https://doi.org/10.18821/0869-2084-2019-64-8-497-502

    • Google Scholar
42. 1. Schmithausen RM
    2. Sib E
    3. Exner M
    4. Hack S
    5. Rösing C
    6. Ciorba P
    7. Bierbaum G
    8. Savin M
    9. Bloomfield SF
    10. Kaase M
    11. Jacobshagen A
    12. Gemein S
    13. Gebel J
    14. Engelhart S
    15. Exner D

    (2019) The washing machine as a reservoir for transmission of Extended-Spectrum-Beta-Lactamase (CTX-M-15)-Producing Klebsiella oxytoca ST201 to newborns

    Applied and Environmental Microbiology 85:e01435-19.

    https://doi.org/10.1128/AEM.01435-19

    • PubMed
    • Google Scholar
43. 1. Schweizer C
    2. Bischoff P
    3. Bender J
    4. Kola A
    5. Gastmeier P
    6. Hummel M
    7. Klefisch FR
    8. Schoenrath F
    9. Frühauf A
    10. Pfeifer Y

    (2019) Plasmid-Mediated transmission of KPC-2 carbapenemase in Enterobacteriaceae in Critically Ill Patients

    Frontiers in Microbiology 10:276.

    https://doi.org/10.3389/fmicb.2019.00276

    • PubMed
    • Google Scholar
44. 1. Seemann T

    (2014) Prokka: rapid prokaryotic genome annotation

    Bioinformatics 30:2068–2069.

    https://doi.org/10.1093/bioinformatics/btu153

    • PubMed
    • Google Scholar
45. 1. Shannon P
    2. Markiel A
    3. Ozier O
    4. Baliga NS
    5. Wang JT
    6. Ramage D
    7. Amin N
    8. Schwikowski B
    9. Ideker T

    (2003) Cytoscape: a software environment for integrated models of biomolecular interaction networks

    Genome Research 13:2498–2504.

    https://doi.org/10.1101/gr.1239303

    • PubMed
    • Google Scholar
46. 1. Sheppard AE
    2. Stoesser N
    3. Wilson DJ
    4. Sebra R
    5. Kasarskis A
    6. Anson LW
    7. Giess A
    8. Pankhurst LJ
    9. Vaughan A
    10. Grim CJ
    11. Cox HL
    12. Yeh AJ
    13. Sifri CD
    14. Walker AS
    15. Peto TE
    16. Crook DW
    17. Mathers AJ

    (2016) Nested Russian Doll-Like Genetic Mobility Drives Rapid Dissemination of the Carbapenem Resistance Gene bla _KPC

    Antimicrobial Agents and Chemotherapy 60:3767–3778.

    https://doi.org/10.1128/AAC.00464-16

    • Google Scholar
47. 1. Stadler T
    2. Meinel D
    3. Aguilar-Bultet L
    4. Huisman JS
    5. Schindler R
    6. Egli A
    7. Seth-Smith HMB
    8. Eichenberger L
    9. Brodmann P
    10. Hübner P
    11. Bagutti C
    12. Tschudin-Sutter S

    (2018) Transmission of ESBL-producing Enterobacteriaceae and their mobile genetic elements-identification of sources by whole genome sequencing: study protocol for an observational study in Switzerland

    BMJ Open 8:e021823.

    https://doi.org/10.1136/bmjopen-2018-021823

    • PubMed
    • Google Scholar
48. 1. Stamatakis A

    (2014) RAxML version 8: a tool for phylogenetic analysis and post-analysis of large phylogenies

    Bioinformatics 30:1312–1313.

    https://doi.org/10.1093/bioinformatics/btu033

    • PubMed
    • Google Scholar
49. 1. Struelens MJ

    (1998) The epidemiology of antimicrobial resistance in hospital acquired infections: problems and possible solutions

    BMJ 317:652–654.

    https://doi.org/10.1136/bmj.317.7159.652

    • PubMed
    • Google Scholar
50. 1. Sullivan MJ
    2. Petty NK
    3. Beatson SA

    (2011) Easyfig: a genome comparison visualizer

    Bioinformatics 27:1009–1010.

    https://doi.org/10.1093/bioinformatics/btr039

    • PubMed
    • Google Scholar
51. 1. Sundermann AJ
    2. Babiker A
    3. Marsh JW
    4. Shutt KA
    5. Mustapha MM
    6. Pasculle AW
    7. Ezeonwuka C
    8. Saul MI
    9. Pacey MP
    10. Van Tyne D
    11. Ayres AM
    12. Cooper VS
    13. Snyder GM
    14. Harrison LH

    (2019a) Outbreak of Vancomycin-resistant Enterococcus faecium in interventional radiology: detection through Whole-genome Sequencing-based surveillance

    Clinical Infectious Diseases 37:ciz666.

    https://doi.org/10.1093/cid/ciz666

    • Google Scholar
52. 1. Sundermann AJ
    2. Miller JK
    3. Marsh JW
    4. Saul MI
    5. Shutt KA
    6. Pacey M
    7. Mustapha MM
    8. Ayres A
    9. Pasculle AW
    10. Chen J
    11. Snyder GM
    12. Dubrawski AW
    13. Harrison LH

    (2019b) Automated data mining of the electronic health record for investigation of healthcare-associated outbreaks

    Infection Control & Hospital Epidemiology 40:314–319.

    https://doi.org/10.1017/ice.2018.343

    • PubMed
    • Google Scholar
53. 1. Valenzuela JK
    2. Thomas L
    3. Partridge SR
    4. van der Reijden T
    5. Dijkshoorn L
    6. Iredell J

    (2007) Horizontal gene transfer in a polyclonal outbreak of carbapenem-resistant Acinetobacter baumannii

    Journal of Clinical Microbiology 45:453–460.

    https://doi.org/10.1128/JCM.01971-06

    • PubMed
    • Google Scholar
54. 1. Ward DV
    2. Hoss AG
    3. Kolde R
    4. van Aggelen HC
    5. Loving J
    6. Smith SA
    7. Mack DA
    8. Kathirvel R
    9. Halperin JA
    10. Buell DJ
    11. Wong BE
    12. Ashworth JL
    13. Fortunato-Habib MM
    14. Xu L
    15. Barton BA
    16. Lazar P
    17. Carmona JJ
    18. Mathew J
    19. Salgo IS
    20. Gross BD
    21. Ellison RT

    (2019) Integration of genomic and clinical data augments surveillance of healthcare-acquired infections

    Infection Control & Hospital Epidemiology 40:649–655.

    https://doi.org/10.1017/ice.2019.75

    • PubMed
    • Google Scholar
55. 1. Wick RR
    2. Judd LM
    3. Gorrie CL
    4. Holt KE

    (2017) Unicycler: resolving bacterial genome assemblies from short and long sequencing reads

    PLOS Computational Biology 13:e1005595.

    https://doi.org/10.1371/journal.pcbi.1005595

    • PubMed
    • Google Scholar
56. 1. Wood DE
    2. Salzberg SL

    (2014) Kraken: ultrafast metagenomic sequence classification using exact alignments

    Genome Biology 15:R46.

    https://doi.org/10.1186/gb-2014-15-3-r46

    • PubMed
    • Google Scholar
57. 1. Zankari E
    2. Hasman H
    3. Cosentino S
    4. Vestergaard M
    5. Rasmussen S
    6. Lund O
    7. Aarestrup FM
    8. Larsen MV

    (2012) Identification of acquired antimicrobial resistance genes

    Journal of Antimicrobial Chemotherapy 67:2640–2644.

    https://doi.org/10.1093/jac/dks261

    • PubMed
    • Google Scholar

Acceptance summary:

This paper shows the process of horizontal gene transfer within a hospital at an unprecedented level of detail and resolution.

Decision letter after peer review:

Thank you for submitting your article "Comprehensive analysis of horizontal gene transfer among multidrug-resistant bacterial pathogens in a single hospital" for consideration by eLife. Your article has been reviewed by three peer reviewers, including Marc Lipsitch as the Reviewing Editor and Reviewer #1, and the evaluation has been overseen by Neil Ferguson as the Senior Editor.

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

Summary:

This paper describes an extensive sequencing project of over 2000 bacterial isolates including the mobile elements from an 18-month period at the University of Pittsburgh hospital. Specifically, authors screened 2173 genomes by using an all-by-all alignment with nucmer to identify shared regions of >5kb and 100% identity. Shared sequences were found in 192 genomes across 11 genera, which were then grouped into 51 clusters of related sequences (ranging from 2-52 genomes in a cluster, with 2-4 genera). Within these clusters, they selected a sample for long-read sequencing, to resolve plasmids, and identified 17 plasmids, with 10 present in >1 sample. They then aligned short reads from all samples to these plasmids, to predict which were present in each. For two patients, authors identified epidemiological links suggesting potential transfer of plasmids within a species, as well as two across-species transfers within the same patients. The identification of plasmid transfer both within and between patients is clinically relevant and highlights the value of incorporating long-read sequencing data in hospital-based surveillance for infection control. Combining these sequencing results with epidemiologic data and enhancing them with hybrid assemblies to understand plasmid transfer, the authors compose a portrait of horizontal gene transfer in a well-defined hospital population that is more extensive than any of which I am aware. As a descriptive study and a data set that will be rich for further analysis, this is a remarkable piece of work which can be made better with some further analysis and improved presentation.

The revisions below are taken from two of the three reviews, which found the manuscript valuable but in need of some revision. For the authors' information, we note a dissenting view from one reviewer, who found the significance of the paper to be less. They wrote: "Notwithstanding the scope of the sampling and numbers of strains examined, there is not much in the way of new information or novel findings. The dissemination of plasmids, mobile elements, antibiotic resistance gene, etc. in hospitals and other settings is a topic of numerous publications, and the degree to which genes can be transferred within and among species, sometimes between distantly related species, is well established." They further noted the uncertainty of the transmission inferences and questioned how generalizable the study would be to other kinds of institutions. These points are noted but the opinion of the reviewing editor is that notwithstanding these caveats, the scale and completeness of the study give it adequate importance to be publishable in eLife, pending revisions.

Essential revisions:

1) At no point in the paper is any justification given for the choices made about the limitation to MDR isolates (or really what that means), the 5000 bp identity requirement, the exclusive focus on inter-genera transfer (while the paper does in fact offer some tantalizing discussion of within-genus transfer), or the very limited information given about clusters after the first 5. Also, the Materials and methods do not describe how the decision to further investigate e.g. epidemiologic links or extent of sequence overlap (pairwise versus fully connected) was made. This manuscript is well-organized but gives a bit of an impression that the sequencing was done, and then some lines of inquiry that seemed inquiry were followed up, until a certain amount of time/effort/results, and a paper was written. This is a perfectly fine thing to do, especially with such interesting material, but it leaves the reader wondering what the full story is. The title says, "comprehensive analysis…among multi-drug resistant pathogens in a single hospital." There is a lot of bioinformatic analysis, but the phrase "comprehensive analysis" would suggest that more things were measured and quantified, and the word comprehensive would suggest that within-genus transfers were also identified and considered. Some specific questions that should be answered if possible are:

– Was there a single definition of an inferred transfer event? Is being in a cluster necessary and sufficient for that inference? Should other definitions be considered?

– Is the inferred amount of HGT between genera high or low or intermediate? Compared to what prior estimates?

– How does the amount of HGT between genera compare to the amount inferred within a genus (perhaps normalized for opportunities to see this)?

– What proportion of inferred events have a plausible path of epidemiologic links?

– What is the impact of the 5kb identity requirement – do you get radically different answers with 3 or 10 as cut-offs?

– What descriptions can be given in quantitative terms of the different patterns of what was conserved within a cluster between C1 and C2, C3? What about all the other clusters?

– What proportion of the inferred events involve mechanisms of horizontal transfer consistent with what we already know (e.g. plasmid transfer, ICE, etc), and which are unexplained by those?

– How do the inferred rates of evolution from the inferred transfer events compare with what we know? The Tn7 comment is tantalizing but is one example – a comprehensive analysis would consider the overall patterns.

– What was the extent of movement of MDR determinants together among inferred events?

Perhaps not all of these can be answered, but to ignore them seems unfortunate in a paper with such a grand title. I don't want to dictate the publication strategy for what will undoubtedly be a series of papers, but a paper in eLife that is called a "Comprehensive Analysis" should not, for example, consider only transfer between genera.

2) Authors state the epidemiology of MGEs in clinical settings requires detailed individual level data, but actually provide nearly no epidemiological data in the current manuscript. I would have expected a table at a minimum outlining the demographics and clinical characteristics of the patients included (N=2173). It would also be helpful to know more about the demographics and clinical characteristics of the patients whose isolates share sequences by cluster (N-192). For example, looking at Figure 1B, I note there are 13 clusters containing Stenotrophomonas – 12 of which are clustered only with Pseudomonas. This would suggest to me that these may be patients with Cystic Fibrosis, as both pathogens are commonly found in the CF lung, but it would be helpful to know this information to better assess the clinical relevance of this work.

3) Authors did long read sequencing on a subset of "representative isolates from the largest clusters" – what do authors mean by “representative” here? Do they just mean they chose a random sample from within each cluster? Please explain.

4) Much of the interpretive material is confusing or questionable.

– The sentence “Taken together, these results indicate that while many of the sequences we identified were shared between related bacterial genera, our approach also identified sequences that were identical in the genomes of distantly related pathogens.” took about four reads before I decided it simply meant there was a lot of sharing among close genera, and some sharing among more distant genera. A more parallel structure to the sentence could clarify (if that is indeed what it means). A reference to Figure 2 could also help.

– Discussion section: "generate biased interpretations of the driving forces": I don't see any interpretation of the driving forces behind HGT (or as I imagine driving forces, behind the success of lineages which have undergone HGT, such as transmissibility or antimicrobial selection pressure or the like) in this paper, and moreover, biased interpretations can be biased only relative to some defined estimand. This seems like unduly vague language, and maybe should be replaced with "incomplete accounts of the extent of HGT" or something similar.

– Discussion paragraph three is somewhat peculiar. It seems to rest on the premise that if an element moves to a new host, it will be selected to change its sequence to adapt to that host, but then undermines that premise with hypotheses 2 and 3. This may be just a matter of wording, but it seems confusing. Maybe sequences are adapted to generic functions (e.g. neutralizing a drug) rather than to the bacterial host. At a minimum the wording should be changed; better would be, instead of giving one example of each, to try to classify the clusters based on these explanations. Again, this is part of the distance between "comprehensive analysis" and the more descriptive tone of the paper.

– "both plasmids…from the same patient were nearly identical to one another, suggesting that they were indeed transferred shortly before the bacteria were isolated" – over what timescale might we expect difference to occur in plasmids and of what magnitude?

– "underscores how quickly MGEs can move" – this makes no sense. MGEs move by for example conjugation which has been measured in the lab as taking minutes. The literal movement is of course fast. I think finding evidence of transfer that is close together in space and time is unremarkable; finding persistence over time is more remarkable.

– Some of the other conclusions seem a bit unsupported by the analyses that are conducted – e.g. "the fact we only observed plasmids in closely related bacterial lineages suggests that they are well-adapted to these lineages, and if they were transmitted to other STs they were likely lost due to instability and/or fitness costs". I would think this could easily be affected by sampling strategy used, with only invasive samples of select species.

– Figure 1D and E just show the proportion of clusters with X gene type or Y AME gene, respectively, but underlying cluster sizes range from 2-52 – shared across all samples. Is there a way authors can standardize by cluster size, as I am not sure a cluster of 2 should have the same weight as a cluster of 52 in these analyses?

– Authors required a minimum of 5kb and 100% identity on nucmer and state these cut-offs were arbitrary (Discussion) – did authors examine any other cut-offs and how do their findings change if these are adjusted?

5) Authors report that they aligned short-reads from all of the isolates to the reference sequences they generated for cluster-containing MGEs (chromosomal or plasmid). They then assessed the coverage (whether this is the% alignment to the reference or depth is unclear – please clarify) to these references to predict whether the respective genome contained the MGE or not. However, I could not find a table showing these results. It would be very helpful to have, for each isolate, the percent of each of these references covered and the median depth of coverage in order to assess the reliability of these results. At a minimum, this should be provided for the plasmid analysis, wherein they found 93 isolates had cluster C1-C5 sequences of 17 plasmids.

6) Transmission analyses – Given the high percent identity between plasmids, the results suggesting transmission of plasmids between patient A and B look quite convincing. However, it would help to have the dates of sample collection for each of these samples. Currently, authors just state patient B's isolate was collected after patient A's. As these are invasive isolates only, it is also is possible that transmission occurred via a colonized intermediary who was not detected due to the study design.

7) A major conclusion authors reach is that some plasmids carrying putative MGEs "were likely inherited vertically as bacteria were transmitted between patients in the hospital." I am a bit surprised not to see any analysis to assess whether the bacterial chromosomes were indeed the same as well as the plasmid given authors have complete Illumina data for these isolates. Authors could easily align reads to a chromosome reference from the respective ST and assess whether this is indeed probable, rather than speculating based on the MGE/plasmid analysis alone.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Systematic detection of horizontal gene transfer across genera among multidrug-resistant bacteria in a single hospital" for further consideration by eLife. Your revised article has been evaluated by Neil Ferguson (Senior Editor) and a Reviewing Editor.

The manuscript has been improved but there are some remaining issues that need to be addressed before acceptance, as outlined below:

Specifically, we have one remaining concern regarding the 5kb threshold for the cluster analysis. The 5kb threshold (which was used in your original analysis), produced 51 clusters. As requested by reviewers, in the revised manuscript the clustering results with >3kb and >10kb are now shown. However, compared to the 5kb threshold, using a 3kb threshold results in 120 clusters, while using a 10kb threshold results in only 16. Despite finding such a wide range in clustering due to this, this has not been discussed anywhere in the current paper – or the implications this would have on the conclusions. It therefore seems like the results are very sensitive to the threshold used, so this should at least be discussed in the paper as a limitation of this work.

Essential revisions:

1) At no point in the paper is any justification given for the choices made about the limitation to MDR isolates (or really what that means), the 5000 bp identity requirement, the exclusive focus on inter-genera transfer (while the paper does in fact offer some tantalizing discussion of within-genus transfer), or the very limited information given about clusters after the first 5. Also, the Materials and methods do not describe how the decision to further investigate e.g. epidemiologic links or extent of sequence overlap (pairwise versus fully connected) was made. This manuscript is well-organized but gives a bit of an impression that the sequencing was done, and then some lines of inquiry that seemed inquiry were followed up, until a certain amount of time/effort/results, and a paper was written. This is a perfectly fine thing to do, especially with such interesting material, but it leaves the reader wondering what the full story is. The title says, "comprehensive analysis…among multi-drug resistant pathogens in a single hospital." There is a lot of bioinformatic analysis, but the phrase "comprehensive analysis" would suggest that more things were measured and quantified, and the word comprehensive would suggest that within-genus transfers were also identified and considered. Some specific questions that should be answered if possible are:

– Was there a single definition of an inferred transfer event? Is being in a cluster necessary and sufficient for that inference? Should other definitions be considered?

– Is the inferred amount of HGT between genera high or low or intermediate? Compared to what prior estimates?

– How does the amount of HGT between genera compare to the amount inferred within a genus (perhaps normalized for opportunities to see this)?

– What proportion of inferred events have a plausible path of epidemiologic links?

– What is the impact of the 5kb identity requirement – do you get radically different answers with 3 or 10 as cut-offs?

– What descriptions can be given in quantitative terms of the different patterns of what was conserved within a cluster between C1 and C2, C3? What about all the other clusters?

– What proportion of the inferred events involve mechanisms of horizontal transfer consistent with what we already know (e.g. plasmid transfer, ICE, etc), and which are unexplained by those?

– How do the inferred rates of evolution from the inferred transfer events compare with what we know? The Tn7 comment is tantalizing but is one example – a comprehensive analysis would consider the overall patterns.

– What was the extent of movement of MDR determinants together among inferred events?

Perhaps not all of these can be answered, but to ignore them seems unfortunate in a paper with such a grand title. I don't want to dictate the publication strategy for what will undoubtedly be a series of papers, but a paper in eLife that is called a "Comprehensive Analysis" should not, for example, consider only transfer between genera.

We appreciate this feedback and agree that the title of our original submission was perhaps misleading. We have modified the title to more accurately reflect what we have actually done in this study. We have added a definition of inferred transfer events and have incorporated a discussion of how little is currently known about plasmid transfer rates in vivo in the Results section. To address these comments more generally, we have deposited all of the Illumina sequence data, as well as the hybrid assembled genomes we generated, into relevant NCBI databases (see Supplementary files 1 and 3). In making the sequence data from our study publicly available, anyone who wishes to will be able to conduct their own analyses (using whatever parameters they choose) on this dataset. Nonetheless, we have repeated our cluster analysis using 3kb and 10kb cut-offs and have added the resulting shared sequence clusters as an additional figure (Figure 1—figure supplement 1). To address the reviewer’s comments about cluster contents and possible mechanisms of transfer, we have added a table describing the gene content of each cluster to the revised manuscript (Supplementary file 2). As we state in the Discussion, we plan to further investigate within-genus transfer in future work. With all of our sequence data publicly available, however, other researchers are free to query this dataset in whatever way(s) they might like.

2) Authors state the epidemiology of MGEs in clinical settings requires detailed individual level data, but actually provide nearly no epidemiological data in the current manuscript. I would have expected a table at a minimum outlining the demographics and clinical characteristics of the patients included (N=2173). It would also be helpful to know more about the demographics and clinical characteristics of the patients whose isolates share sequences by cluster (N-192). For example, looking at Figure 1B, I note there are 13 clusters containing Stenotrophomonas – 12 of which are clustered only with Pseudomonas. This would suggest to me that these may be patients with Cystic Fibrosis, as both pathogens are commonly found in the CF lung, but it would be helpful to know this information to better assess the clinical relevance of this work.

The is an excellent point. We have now included a table that summarizes relevant demographic and clinical information from the patients in the study (Table 1).

3) Authors did long read sequencing on a subset of "representative isolates from the largest clusters" – what do authors mean by “representative” here? Do they just mean they chose a random sample from within each cluster? Please explain.

We selected isolates for long-read sequencing by considering the length of the shared sequence (i.e. the isolate with the longest sequence was selected), as well as culture date (i.e. selecting the earliest isolate). We have included this information in the revised manuscript Results section.

4) Much of the interpretive material is confusing or questionable.

– The sentence “Taken together, these results indicate that while many of the sequences we identified were shared between related bacterial genera, our approach also identified sequences that were identical in the genomes of distantly related pathogens.” took about four reads before I decided it simply meant there was a lot of sharing among close genera, and some sharing among more distant genera. A more parallel structure to the sentence could clarify (if that is indeed what it means). A reference to Figure 2 could also help.

We apologize for the confusion. This sentence has been removed from the revised manuscript.

– Discussion section: "generate biased interpretations of the driving forces": I don't see any interpretation of the driving forces behind HGT (or as I imagine driving forces, behind the success of lineages which have undergone HGT, such as transmissibility or antimicrobial selection pressure or the like) in this paper, and moreover, biased interpretations can be biased only relative to some defined estimand. This seems like unduly vague language, and maybe should be replaced with "incomplete accounts of the extent of HGT" or something similar.

We agree and have modified this sentence accordingly.

– Discussion paragraph three is somewhat peculiar. It seems to rest on the premise that if an element moves to a new host, it will be selected to change its sequence to adapt to that host, but then undermines that premise with hypotheses 2 and 3. This may be just a matter of wording, but it seems confusing. Maybe sequences are adapted to generic functions (e.g. neutralizing a drug) rather than to the bacterial host. At a minimum the wording should be changed; better would be, instead of giving one example of each, to try to classify the clusters based on these explanations. Again, this is part of the distance between "comprehensive analysis" and the more descriptive tone of the paper.

We thank the reviewer for this helpful feedback. The three possible causes that we listed are hypotheses that might help explain our observations, and for which we have some (very limited) evidence in our dataset. We chose to offer some speculation regarding what we feel is the strongest example supporting each hypothesis; formally testing each one, especially in a dataset of this size, would require significant additional work that we feel would be better suited to future studies. We have nonetheless modified this section of the manuscript accordingly.

– "both plasmids…from the same patient were nearly identical to one another, suggesting that they were indeed transferred shortly before the bacteria were isolated" – over what timescale might we expect difference to occur in plasmids and of what magnitude?

This is an important point to consider. Unfortunately, careful estimates of mutation rates among mobile elements are currently lacking. We have revised this section of the manuscript accordingly.

– "underscores how quickly MGEs can move" – this makes no sense. MGEs move by for example conjugation which has been measured in the lab as taking minutes. The literal movement is of course fast. I think finding evidence of transfer that is close together in space and time is unremarkable; finding persistence over time is more remarkable.

This is a fair point and we have modified this sentence accordingly.

– Some of the other conclusions seem a bit unsupported by the analyses that are conducted – e.g. "the fact we only observed plasmids in closely related bacterial lineages suggests that they are well-adapted to these lineages, and if they were transmitted to other STs they were likely lost due to instability and/or fitness costs". I would think this could easily be affected by sampling strategy used, with only invasive samples of select species.

This is also a fair point, and we have removed this sentence from the revised manuscript.

– Figure 1D and E just show the proportion of clusters with X gene type or Y AME gene, respectively, but underlying cluster sizes range from 2-52 – shared across all samples. Is there a way authors can standardize by cluster size, as I am not sure a cluster of 2 should have the same weight as a cluster of 52 in these analyses?

We agree that cluster size might affect how we interpret the contents of each cluster. We now include a table that summarizes the number of isolates and genetic contents of each cluster (Supplementary file 2), which will allow readers to access this information directly.

– Authors required a minimum of 5Kb and 100% identity on nucmer and state these cut-offs were arbitrary (Discussion) – did authors examine any other cut-offs and how do their findings change if these are adjusted?

In line with this and the comment above, we repeated our cluster detection using 3kb and 10kb cut-offs and have included the results in the revised manuscript in the Results section (also see Figure 1—figure supplement 1). In the future we plan to explore different identity cut-offs as well.

5) Authors report that they aligned short-reads from all of the isolates to the reference sequences they generated for cluster-containing MGEs (chromosomal or plasmid). They then assessed the coverage (whether this is the% alignment to the reference or depth is unclear – please clarify) to these references to predict whether the respective genome contained the MGE or not. However, I could not find a table showing these results. It would be very helpful to have, for each isolate, the percent of each of these references covered and the median depth of coverage in order to assess the reliability of these results. At a minimum, this should be provided for the plasmid analysis, wherein they found 93 isolates had cluster C1-C5 sequences of 17 plasmids.

We thank the reviewer for this feedback. To assess reference sequence presence, we aligned contigs from Illumina assemblies to each reference sequence, rather than raw reads. Because of this there was no cut-off used for read depth. We have included a table summarizing the coverage results from our contig-mapping analysis (Supplementary file 4), and both the raw read data for all study isolates and the reference plasmid sequences have been posted to SRA/GenBank, so that others can conduct analyses with this data.

6) Transmission analyses – Given the high percent identity between plasmids, the results suggesting transmission of plasmids between patient A and B look quite convincing. However, it would help to have the dates of sample collection for each of these samples. Currently, authors just state patient B's isolate was collected after patient A's. As these are invasive isolates only, it is also is possible that transmission occurred via a colonized intermediary who was not detected due to the study design.

We agree that providing additional information about the plasmid transferred between Patients A and B would be helpful. We have added information about the time separating the collection of their isolates in the Results section, as well as additional speculation that transfer could have happened during colonization, which we did not assess, in the Discussion section.

7) A major conclusion authors reach is that some plasmids carrying putative MGEs "were likely inherited vertically as bacteria were transmitted between patients in the hospital." I am a bit surprised not to see any analysis to assess whether the bacterial chromosomes were indeed the same as well as the plasmid given authors have complete Illumina data for these isolates. Authors could easily align reads to a chromosome reference from the respective ST and assess whether this is indeed probable, rather than speculating based on the MGE/plasmid analysis alone.

This is a valid point. We have looked more closely at the K. pneumoniae ST258 and E. coli ST131 genomes in our dataset and found that in both cases, the isolates carrying these plasmids were more closely related to each other than to other isolates lacking the plasmids. This finding supports our conclusion that the plasmids were likely transmitted along with the bacteria carrying them and has been added to the revised manuscript in the Results section (see also Figure 5—figure supplement 1).

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

The manuscript has been improved but there are some remaining issues that need to be addressed before acceptance, as outlined below:

Specifically, we have one remaining concern regarding the 5kb threshold for the cluster analysis. The 5kb threshold (which was used in your original analysis), produced 51 clusters. As requested by reviewers, in the revised manuscript the clustering results with >3kb and >10kb are now shown. However, compared to the 5kb threshold, using a 3kb threshold results in 120 clusters, while using a 10kb threshold results in only 16. Despite finding such a wide range in clustering due to this, this has not been discussed anywhere in the current paper – or the implications this would have on the conclusions. It therefore seems like the results are very sensitive to the threshold used, so this should at least be discussed in the paper as a limitation of this work.

This is an important point. Cluster number will also vary based on the identity threshold used – if we lowered the nucleotide identity threshold below 100% we would almost certainly detect more (and bigger) clusters at all length cut-offs. We agree that this is an additional limitation that warrants attention and have incorporated this idea into both the Results and Discussion sections in the revised manuscript.

Author details

1. Daniel R Evans

   1. Division of Infectious Diseases, University of Pittsburgh School of Medicine, Pittsburgh, United States
   2. Department of Infectious Diseases and Microbiology, University of Pittsburgh Graduate School of Public Health, Pittsburgh, United States

   Contribution

   Conceptualization, Resources, Data curation, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing - original draft, Writing - review and editing

   Competing interests

   No competing interests declared

2. Marissa P Griffith

   Microbial Genomic Epidemiology Laboratory, Infectious Diseases Epidemiology Research Unit, University of Pittsburgh School of Medicine and Graduate School of Public Health, Pittsburgh, United States

   Contribution

   Data curation, Software, Formal analysis, Validation, Investigation, Methodology, Writing - review and editing

   Competing interests

   No competing interests declared

3. Alexander J Sundermann

   Microbial Genomic Epidemiology Laboratory, Infectious Diseases Epidemiology Research Unit, University of Pittsburgh School of Medicine and Graduate School of Public Health, Pittsburgh, United States

   Contribution

   Formal analysis, Investigation, Writing - review and editing

   Competing interests

   No competing interests declared

4. Kathleen A Shutt

   Microbial Genomic Epidemiology Laboratory, Infectious Diseases Epidemiology Research Unit, University of Pittsburgh School of Medicine and Graduate School of Public Health, Pittsburgh, United States

   Contribution

   Data curation, Formal analysis, Writing - review and editing

   Competing interests

   No competing interests declared

5. Melissa I Saul

   Department of Medicine, University of Pittsburgh School of Medicine, Pittsburgh, United States

   Contribution

   Data curation, Formal analysis, Writing - review and editing

   Competing interests

   No competing interests declared

6. Mustapha M Mustapha

   Microbial Genomic Epidemiology Laboratory, Infectious Diseases Epidemiology Research Unit, University of Pittsburgh School of Medicine and Graduate School of Public Health, Pittsburgh, United States

   Contribution

   Conceptualization, Data curation, Formal analysis, Investigation, Writing - review and editing

   Competing interests

   No competing interests declared

7. Jane W Marsh

   Microbial Genomic Epidemiology Laboratory, Infectious Diseases Epidemiology Research Unit, University of Pittsburgh School of Medicine and Graduate School of Public Health, Pittsburgh, United States

   Contribution

   Conceptualization, Investigation, Project administration, Writing - review and editing

   Competing interests

   No competing interests declared

8. Vaughn S Cooper

   Department of Microbiology and Molecular Genetics, and Center for Evolutionary Biology and Medicine, University of Pittsburgh, Pittsburgh, United States

   Contribution

   Conceptualization, Investigation, Writing - review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0001-7726-0765

9. Lee H Harrison

   Microbial Genomic Epidemiology Laboratory, Infectious Diseases Epidemiology Research Unit, University of Pittsburgh School of Medicine and Graduate School of Public Health, Pittsburgh, United States

   Contribution

   Conceptualization, Supervision, Funding acquisition, Investigation, Project administration, Writing - review and editing

   Competing interests

   No competing interests declared

10. Daria Van Tyne

    Division of Infectious Diseases, University of Pittsburgh School of Medicine, Pittsburgh, United States

    Contribution

    Conceptualization, Data curation, Formal analysis, Supervision, Funding acquisition, Investigation, Visualization, Methodology, Writing - original draft, Project administration, Writing - review and editing

    For correspondence

    VANTYNE@pitt.edu

    Competing interests

    No competing interests declared

    "This ORCID iD identifies the author of this article:" 0000-0001-7284-0103

• 9,302

  Page views

• 993

  Downloads

• 40

  Citations

Article citation count generated by polling the highest count across the following sources: Scopus, Crossref, PubMed Central.