Vet Dermatol 2021; 32: 654–663 DOI: 10.1111/vde.13040 Whole genome sequencing and de novo assembly of Staphylococcus pseudintermedius: a pangenome approach to unravelling pathogenesis of canine pyoderma Lluıs Ferrer* , Rocıo Garcıa-Fonticoba* rez† , Joaquim Vin , Daniel Pe ~ es†,‡ , Norma Fa bregas‡ , Sergi Madron ~ ero† , Gabriele Meroni§ , Piera A. Martino§ ,    Sofıa Martınez¶ , M. Laura Mate**, Sergio Sa nchez-Bruni** , Anna Cusco ‡ , Lourdes Migura-Garcıa†† and Olga Francino† *Departament de Medicina i Cirurgia Animals, Universitat Auto noma de Barcelona, Barcelona, Spain tica Molecular, Universitat Auto †Servei Veterinari de Gene noma de Barcelona, Barcelona, Spain ‡Vetgenomics, PRUAB, Bellaterra, Barcelona, Spain §Department of Biomedical, Surgical and Dental Sciences – One Health Unit, Via Pascal 36, Milano, 20133, Italy ¶Hospital Escuela de Pequen ~os Animales (HEPA), Centro de Investigacio n Veterinaria de Tandil (CIVETAN), UNCPBA-CONICET-CICPBA, Grupo de Medicina Veterinaria Traslacional (MEVET), Facultad de Ciencias Veterinarias, Universidad Nacional del Centro de la Provincia de Buenos Aires (FCV-UNCPBA), Tandil, Argentina **Laboratorio de Farmacologıa, Centro de Investigacion Veterinaria de Tandil (CIVETAN), UNCPBA-CONICET-CICPBA, Facultad de Ciencias Veteri- narias, Universidad Nacional del Centro (FCV-UNCPBA), Tandil, Argentina ††Centre de Recerca en Sanitat Animal, IRTA, Bellaterra, Barcelona, Spain noma de Barcelona, 08193 Bellaterra, Barcelona, Correspondence: Lluıs Ferrer, Department de Medicina I Cirurgia Animals, Universitat Auto Spain. E-mail: lluis.ferrer@uab.cat Background – Staphylococcus pseudintermedius is the main aetiological agent of canine pyoderma. Whole gen- ome sequencing is the most comprehensive way of obtaining relevant genomic information about micro- organisms. Hypothesis/Objectives – Oxford Nanopore technology enables quality sequencing and de novo assembly of the whole genome of S. pseudintermedius. Whole genome analysis of S. pseudintermedius may help to better understand the pathogenesis of canine pyodermas. Methods and materials – Twenty-two strains of S. pseudintermedius isolated from the skin of five healthy dogs and 33 strains isolated from skin of 33 dogs with pyoderma were analysed. DNA was extracted and sequen- ced using Oxford Nanopore MinION, a new technology that delivers longer reads in a hand-held device. The pan- genome was analysed and visualised with ANVI’O 6.1. Results – Nanopore technology allowed the sequencing and de novo assembly of the genomes of 55 S. pseu- dintermedius strains isolated from healthy dogs and from dogs with pyoderma. The average genome size of S. pseudintermedius was 2.62 Mbp, with 48% being core genome. Pyoderma isolates contained a higher num- ber of antimicrobial resistance genes, yet the total number of virulence factors genes did not change between isolates from healthy dogs and from dogs with pyoderma. Genomes of meticillin-resistant S. pseudintermedius (MRSP) strains were larger than those of meticillin-susceptible (MSSP) strains (2.80 Mbp versus 2.59 Mbp), as a consequence of a greater presence of antimicrobial resistance genes, phages and prophages. Conclusions and clinical importance – This technique allows much more precise and easier characterisation of canine S. pseudintermedius populations and may lead to a better understanding of the pathogenesis of canine pyodermas. Accepted 23 September 2021 Sources of Funding: Research grant RTI2018-101991-B-100 from the Ministerio de Ciencia e Innovacio  n, Spain; Industrial Doctorate Pro- gram grant 2017DI037 AGAUR, Generalitat de Catalunya, Spain. Ministerio de Ciencia e Innovacio  n, Spain: Torres Quevedo Project PTQ2018-009961. Conflicts of Interest: L.F. has received unrelated honoraria for lecturing from Zoetis, Bayer, LETI, and Affinity Petcare. O.F. and A.C. have received unrelated honoraria for lecturing from Oxford Nanopore Technologies. This study was presented as a supporting original study at the 9th World Congress of Veterinary Dermatology, 20–24 October 2020, Syd- ney, Australia. 654 © 2021 The Authors. Veterinary Dermatology published by John Wiley & Sons Ltd on behalf of ESVD and ACVD., 32, 654–663. This is an open access article under the terms of the Creative Commons Attribution NonCommercial License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited and is not used for commercial purposes. Whole-genome-sequencing of S. pseudintermedius Introduction the genome of S. pseudintermedius and detect differ- ences that could explain the change from a commensal Staphylococcus pseudintermedius is a component of the micro-organism to a pathogen. dog skin microbiota1,2 and the main causative agent of pyoderma in this species.3,4 A study using pulse-field gel electrophoresis (PFGE) of dogs with pyoderma concluded Materials and methods that S. pseudintermedius isolated from skin lesions Bacterial cultures (pustules) were identical to the S. pseudintermedius iso- The present study was carried out on 22 cultures of S. pseudinter- lated from nonlesional sites of the same dog.5 This find- medius isolated from the skin of six healthy dogs and 33 cultures ing indicated that the S. pseudintermedius causing isolated from 33 dogs diagnosed with pyoderma. In the case of the skin infections very likely originated from commensal healthy dogs, the samples were obtained by rubbing a sterile swab S. pseudintermedius populations in the canine skin. Thus, on the perioral or abdominal skin for 15 s. In the dogs with pyo- the current paradigm indicates that infection usually derma, samples were obtained, whenever possible, from the con- tent of the pustules. In three cases where no obvious pustules arises when the skin and mucosal barriers are altered by were present, the samples were obtained from epidermal col- predisposing factors such as atopic dermatitis (AD), medi- larettes, as described in the literature.10 The samples were cultured cal and surgical procedures, and/or immunosuppressive in blood agar at 37°C and incubated in aerobiosis for 24 h or 48 h, disorders.3 Therefore, S. pseudintermedius is considered depending on the visual growth on the plate. Colonies with charac- to be an opportunistic pathogen. Nevertheless, the mech- teristic morphology of S. pseudintermedius were subcultured and anisms by which a commensal micro-organism is trans- stored in brain heart infusion (BHI) broth with 20% glycerol at – formed into a pathogen are poorly understood. 80°C for further studies. Before the DNA extraction procedure, cul- tures were plated in blood agar and seeded in 3 mL BHI broth at The increasing development of antibiotic resistance in 37°C for 24 h in aerobiosis. S. pseudintermedius populations and especially meticillin resistance (MR), which confers resistance to all beta- DNA extraction and sequencing lactam antibiotics, is a very serious problem worldwide.4 Staphylococcus pseudintermedius DNA from the BHI broth cultures The reduction and rational use of antibiotics in veterinary was extracted with a Zymo BIOMICS DNA Miniprep Kit (Zymo medicine is one of the main strategies to reduce bacterial Research; Irvine, CA, USA). DNA quality and quantity were deter- resistance and one that is advocated by health organisa- mined using a Nanodrop 2000 Spectrophotometer and Qubit dsDNA tions. Understanding the pathogenic mechanisms of pyo- BR Assay Kit (Fisher Scientific S.L.; Madrid, Spain). The sequencing derma could help prevent the development of pyoderma libraries were prepared using 200 400 ng DNA which were sub- jected to transposase fragmentation using the Rapid Barcoding and help develop new therapies, which could reduce the Sequencing kit (SQK-RBK004, Oxford Nanopore Technologies; use of antibiotics. Nevertheless, to date, it has been chal- Oxford Science Park, UK), and 12 barcoded samples were loaded in a lenging to find a research strategy to learn more about MinION FLO-MIN106 v9.4.1 flow cell (Oxford Nanopore Technolo- how pathogenic S. pseudintermedius populations origi- gies) and sequenced in a MinION Mk1B. The fast5 files were base- nate and the differences – if any – between commensal called with Guppy 4.0.11 (Oxford Nanopore Technologies) with high and pathogenic S. pseudintermedius populations. accuracy base-calling mode, demultiplexed and with adapters trimmed. Reads with a quality score <7 were discarded. The recent development of next-generation sequenc- ing (NGS) techniques that allow relatively easy and eco- nomical mass sequencing of genomes has opened up a Assembly and visualisation of the genomes NANOPLOT 1.27 (https://github.com/wdecoster/NanoPlot) was used to new pathway for studying infectious diseases. In particu- obtain the run summary statistics.11 Sequences corresponding to lar, the sequencing of the complete genome of micro- S. pseudintermedius after taxonomy assignment using What’s in my organisms allows a very precise genotypic characterisa- pot (WIMP) workflow from the EPI2ME platform12 were de novo tion, identifying, for example, factors of virulence or assembled using FLYE 2.7.1.13 MINIMAP 2.17 was used to align the antimicrobial resistance and allowing the comparison of assembled sequences to the raw data files.14 The resulting contigs different strains and isolates. The release of Oxford Nano- then were first polished with the graph-based correction method in RACON 1.4.13 (https://github.com/lbcb-sci/racon), followed by the pore Technologies’ MinION in 2014 generated much neural-network based correction method used by MEDAKA 1.0.3 excitement in the genomics community by offering porta- (https://nanoporetech.github.io/medaka/). bility (it measures approximately 3 cm x 10 cm), speed Genome completeness was assessed with CHECKM 1.1.1.15 CIR- and the capability of producing reads of virtually any CLATOR 1.5.5 was used to identify the origin and the reverse comple- length.6 The fact that it can work with longer reads facili- mentary sequences.16 Genomes were annotated with NCBI tates the assembly of genomes and the characterisation, Prokaryotic Genome Annotation Pipeline (PGAP)17 (after uploading to location and genomic context of virulence and resistance NCBI; Bioproject PRJNA685966), and the total numbers of coding sequences, rRNA and tRNA were determined. Comparative genome genes.7 So even though it has a higher error rate than the identity was assessed with FASTANI.18 Illumina system, it has quickly become very useful for ANVI’O 6.2 (https://merenlab.org/2016/11/08/pangenomics-v2/) rapid clinical responses and sequencing in the field.8,9 allows comparison of shared genes.19 It was used to determine the The objective of this study was two-fold. First, to deter- core genome and accessory genome of S. pseudintermedius iso- mine whether the Nanopore technology allows a lates. sequencing and de novo assembly of the whole genome A custom phylogenetic tree was built with PATRIC20 from the 55 of S. pseudintermedius. Second, in the event that the de novo assembled genomes and 31 complete genomes available in NCBI as of May 2021, with 100 single-copy PATRIC PGFams result was positive, to analyse and compare the genome selected by the codon tree method as homology groups (Fig. 1). The of different S. pseudintermedius strains isolated from the aligned proteins and coding DNA from single-copy genes were anal- skin of healthy dogs and from skin lesions of dogs with ysed with RAxML,21 as included in PATRIC. The resulting Newick file pyoderma. We aimed to reach a better understanding of was viewed in FIGTREE v1.4.4 (https://github.com/rambaut/figtree). © 2021 The Authors. Veterinary Dermatology published by John Wiley & Sons Ltd on behalf of ESVD and ACVD., 32, 654–663. 655 Ferrer et al. Staphylococcus pseudintermedius E140 Staphylococcus pseudintermedius strain G3C4 Staphylococcus pseudintermedius strain DHSP041 Staphylococcus pseudintermedius strain DHSP045 Staphylococcus pseudintermedius strain DHSP044 Staphylococcus pseudintermedius strain DSP036 Staphylococcus pseudintermedius strain DHSP043 Staphylococcus pseudintermedius strain DHSP042 Staphylococcus pseudintermedius strain DG089 Staphylococcus pseudintermedius strain DG063 Staphylococcus pseudintermedius strain DG062 Staphylococcus pseudintermedius strain DG094 Staphylococcus pseudintermedius strain DG067 Staphylococcus pseudintermedius strain DG072 Staphylococcus pseudintermedius strain DG071 Staphylococcus pseudintermedius strain DG077 Staphylococcus pseudintermedius strain DG066 Staphylococcus pseudintermedius strain DSP021 Staphylococcus pseudintermedius strain DSP020 Staphylococcus pseudintermedius strain DSP025 Staphylococcus pseudintermedius strain DG078 Staphylococcus pseudintermedius strain SP_9261-1A Staphylococcus pseudintermedius strain DHSP046 Staphylococcus pseudintermedius strain SP_11306-1A Staphylococcus pseudintermedius strain SP_11306-4A Staphylococcus pseudintermedius strain HSP141 Staphylococcus pseudintermedius strain HSP082 Staphylococcus pseudintermedius strain HSP081 Staphylococcus pseudintermedius strain VB16 Staphylococcus pseudintermedius strain AK9 Staphylococcus pseudintermedius strain 157588 Staphylococcus pseudintermedius strain KCTC 43135 Staphylococcus pseudintermedius strain KCTC 43134 Staphylococcus pseudintermedius strain KCTC 43136 Staphylococcus pseudintermedius ED99 Staphylococcus pseudintermedius strain SP_11304-1A Staphylococcus pseudintermedius strain SP_11304-2A Staphylococcus pseudintermedius strain DG059 Staphylococcus pseudintermedius strain HSP143 Staphylococcus pseudintermedius strain DSP030 Staphylococcus pseudintermedius strain DSP035 Staphylococcus pseudintermedius strain DG050 Staphylococcus pseudintermedius strain DSP032 Staphylococcus pseudintermedius strain HSP140 Staphylococcus pseudintermedius strain HSP142 Staphylococcus pseudintermedius strain HSP137 Staphylococcus pseudintermedius strain HSP118 Staphylococcus pseudintermedius strain HSP136 Staphylococcus pseudintermedius strain HSP134 Staphylococcus pseudintermedius strain HSP132 Staphylococcus pseudintermedius strain HSP138 Staphylococcus pseudintermedius strain HSP135 Staphylococcus pseudintermedius strain HSP079 Staphylococcus pseudintermedius strain HSP080 Staphylococcus pseudintermedius strain DSP027 Staphylococcus pseudintermedius strain DG081 Staphylococcus pseudintermedius strain HSP127 Staphylococcus pseudintermedius strain DSP024 Staphylococcus pseudintermedius strain FDAARGOS_930 strain Not applicable Staphylococcus pseudintermedius strain DSP026 Staphylococcus pseudintermedius strain AI14 Staphylococcus pseudintermedius strain AP20 Staphylococcus pseudintermedius strain 49_44 Staphylococcus pseudintermedius strain 53_88 Staphylococcus pseudintermedius strain SP_11304-3A Staphylococcus pseudintermedius strain DSP034 Staphylococcus pseudintermedius strain 53_60 Staphylococcus pseudintermedius strain 5912 Staphylococcus pseudintermedius strain DG040 Staphylococcus pseudintermedius strain 51_92 Staphylococcus pseudintermedius strain SP79 Staphylococcus pseudintermedius strain DSP028 Staphylococcus pseudintermedius strain DSP022 Staphylococcus pseudintermedius strain DG099 Staphylococcus pseudintermedius strain HSP125 Staphylococcus pseudintermedius strain HSP093 Staphylococcus pseudintermedius strain HSP095 Staphylococcus pseudintermedius strain HSP096 Staphylococcus pseudintermedius strain HSP097 Staphylococcus pseudintermedius strain HSP094 Staphylococcus pseudintermedius strain DG076 Staphylococcus pseudintermedius strain DG082 Staphylococcus pseudintermedius strain DG093 Staphylococcus pseudintermedius strain DSP029 Staphylococcus pseudintermedius strain 063228 Staphylococcus pseudintermedius strain NA45 Staphylococcus pseudintermedius HKU10-03 Staphylococcus pseudintermedius strain cdc 18-1182 Staphylococcus pseudintermedius strain Z0118SP0108 Staphylococcus pseudintermedius strain VB88 Staphylococcus pseudintermedius strain AH18 Staphylococcus pseudintermedius strain FDAARGOS_1073 strain Not applicable Staphylococcus pseudintermedius strain FDAARGOS_1072 strain Not applicable Staphylococcus pseudintermedius strain DG064 Staphylococcus pseudintermedius strain 6109 Staphylococcus pseudintermedius strain 081661 Figure 1. Phylogenetic tree built with PATRIC from 31 complete genomes available in NCBI and the 55 de novo assembled genomes from this study. Thirty-three strains were isolated from dogs with pyoderma (strains with codes DG and DSP) and 22 from the skin of healthy dogs (strains with code HSP)(full data for each strain can be seen in Table 1). Red boxes indicate meticillin-resistant Staphylococcus pseudintermedius (MRSP) clus- ters. Multilocus sequence type, antibiotic-resistance is version 1. Full data from the whole-genome assemblies also have genes, virulence factors and bacteriophages been published elsewhere.28 Multilocus sequence types (MLST) were assigned with MLST 2.0 software and its database 2.0.0.22 Antibiotic resistance genes were identified with ABRICATE 0.8.1323 (https://github.com/tseemann/ Results abricate) with the CARD database.24 Plasmids (replicons) were identi- DNA extraction and purification from BHI broth cultures fied with PLASMIDFINDER 2.1.25 A custom database also was created to developed as expected. Sequencing libraries were pre- analyse the virulence factors (SPVFDB), containing 58 genes encod- pared with 12 isolates per library with the Rapid Barcod- ing for virulence factors that include exfoliative toxins, enterotoxins, leukocidins, pore-forming proteins and intercellular adhesion pro- ing kit (12-plex barcode libraries) and sequenced in the teins. Subsequently, the results were filtered by genes with identity Nanopore MinION Mk1B for 24 h. The mean size of the and coverage ≥90%. PHIGARO 2.2.626 and VIRSORTER 1.0.627 were used reads was 2,600 bp. After quality assessment of the to identify phage and bacteriophage sequences within the genomes. reads, all genomes were assembled, achieving complete- ness >96.7% and considered high-quality (completeness Data availability >90%, contamination <5%) for further analyses. Mean The whole-genome assemblies were deposited at GenBank reposi- sequencing coverage was 9249.18. Table 1 presents the tory under BIOPROJECT PRJNA685966 and with the accession most relevant genomic data of the 55 isolates of numbers CP066702–066718, CP066884, CP066885 and JAENBQ0 00000000–JAENDF000000000. The version described in this paper S. pseudintermedius according to the origin of the 656 © 2021 The Authors. Veterinary Dermatology published by John Wiley & Sons Ltd on behalf of ESVD and ACVD., 32, 654–663. Whole-genome-sequencing of S. pseudintermedius Table 1. Results of nanopore sequencing and de novo assembly of the 55 Staphylococcus pseudintermedius isolates, including the following parameters: health status, isolate, country of origin of isolate, genome size, %GC, multilocus sequence types (MLST), meticillin-resistance (MRSP) and SSC-mec type cassette, number of phages and prophages, virulence factor and antimicrobial resistance genes identified, and multidrug- resistance genotype. Meticillin Virulence Antimicrobial Multidrug Healthy skin/ Origin of Genome resistance Phages & factor resistance resistant pyoderma Isolate sample size (bp) GC % MLST (mecA) SCCmec Type prophages genes genes genotype Pyoderma D_G062 Italy 2991046 37.4 71 MRSP SCCmec_type_II-III 9 41 12 yes Pyoderma D_G063 Italy 2981523 37.4 71 MRSP SCCmec_type_II-III 9 41 11 yes Pyoderma D_G064 Italy 2895060 37.5 71 MRSP SCCmec_type_II-III 7 40 10 yes Pyoderma D_G066 Italy 2808032 37.5 71 MRSP SCCmec_type_II-III 5 41 11 yes Pyoderma D_G067 Italy 2896399 37.5 71 MRSP SCCmec_type_II-III 8 40 10 yes Pyoderma D_G071 Italy 2807986 37.5 71 MRSP SCCmec_type_II-III 5 42 11 yes Pyoderma D_G072 Italy 2837133 37.4 71 MRSP SCCmec_type_II-III 6 41 10 yes Pyoderma D_G077 Italy 2791496 37.5 71 MRSP SCCmec_type_II-III 6 42 10 yes Pyoderma D_G078 Italy 2799396 37.4 71 MRSP SCCmec_type_II-III 5 41 11 yes Pyoderma D_G089 Italy 2797937 37.5 71 MRSP SCCmec_type_II-III 5 41 11 yes Pyoderma D_G094 Italy 2849103 37.4 71 MRSP SCCmec_type_II-III 5 39 11 yes Pyoderma D_SP020 Spain 2793830 37.5 71 MRSP SCCmec_type_II-III 5 42 10 yes Pyoderma D_SP021 Spain 2795724 37.5 71 MRSP SCCmec_type_II-III 5 41 10 yes Pyoderma D_SP025 Spain 2805515 37.5 71 MRSP SCCmec_type_II-III 6 41 10 yes Pyoderma D_SP036 Spain 2927015 37.3 71 MRSP SCCmec_type_II-III 5 41 10 yes Pyoderma D_G076 Italy 2780144 37.2 258 MRSP SCCmec_type_IVg 1 39 8 yes Pyoderma D_G082 Italy 2626557 37.6 258 MRSP SCCmec_type_IVg 1 39 7 yes Pyoderma D_G093 Italy 2648891 37.6 258 MRSP SCCmec_type_IVg 2 39 9 yes Pyoderma D_SP029 Spain 2723805 37.4 258 MRSP SCCmec_type_IVg 1 40 8 yes Pyoderma D_G081 Italy 2694747 37.6 301 MRSP SCCmec_type_IVg 2 38 9 yes Pyoderma D_SP032 Argentina 2670199 37.5 1631 MRSP none 1 43 9 yes Pyoderma D_SP022 Spain 2767901 37.2 Unknown MRSP none 3 41 9 yes Pyoderma D_G050 Italy 2637713 37.6 Unknown MSSP 2 42 8 yes Pyoderma D_G099 Italy 2623014 37.6 Unknown MSSP 1 45 8 yes Pyoderma D_SP035 Argentina 2642865 37.6 Unknown MSSP 2 43 6 yes Pyoderma D_SP028 Spain 2575420 37.6 Unknown MSSP 1 42 6 yes Pyoderma D_SP024 Spain 2762026 37.5 611 MSSP 3 42 4 yes Pyoderma D_SP026 Spain 2567628 37.6 503 MSSP 1 42 1 no Pyoderma D_SP034 Argentina 2550368 37.6 1827 MSSP 0 44 1 no Pyoderma D_SP027 Spain 2717194 37.3 Unknown MSSP 2 44 1 no Pyoderma D_SP030 Argentina 2612059 37.6 Unknown MSSP 1 41 1 no Pyoderma D_G059 Italy 2573568 37.6 Unknown MSSP 1 40 1 no Pyoderma D_G040 Italy 2564892 37.7 Unknown MSSP 1 43 1 no Healthy skin H_SP141 Spain 2622529 37.6 257 MSSP 2 42 2 no Healthy skin H_SP125 Spain 2551473 37.6 1061 MSSP 1 46 1 no Healthy skin H_SP118 Spain 2512855 37.7 1248 MSSP 0 42 1 no Healthy skin H_SP079 Spain 2585691 37.5 Unknown MSSP 0 43 3 yes Healthy skin H_SP080 Spain 2585570 37.5 Unknown MSSP 0 43 3 yes Healthy skin H_SP081 Spain 2621254 37.7 Unknown MSSP 3 40 1 no Healthy skin H_SP082 Spain 2620663 37.7 Unknown MSSP 3 40 1 no Healthy skin H_SP093 Spain 2590335 37.5 Unknown MSSP 2 42 1 no Healthy skin H_SP094 Spain 2570595 37.6 Unknown MSSP 2 42 1 no Healthy skin H_SP095 Spain 2575879 37.6 Unknown MSSP 2 42 1 no Healthy skin H_SP096 Spain 2578330 37.6 Unknown MSSP 2 42 1 no Healthy skin H_SP097 Spain 2575223 37.6 Unknown MSSP 2 42 1 no Healthy skin H_SP127 Spain 2690618 37.5 Unknown MSSP 2 43 2 no Healthy skin H_SP132 Spain 2515164 37.7 Unknown MSSP 0 42 1 no Healthy skin H_SP134 Spain 2514594 37.7 Unknown MSSP 0 42 1 no Healthy skin H_SP135 Spain 2512727 37.7 Unknown MSSP 0 42 1 no Healthy skin H_SP136 Spain 2512726 37.7 Unknown MSSP 0 42 1 no Healthy skin H_SP137 Spain 2512757 37.7 Unknown MSSP 0 42 1 no Healthy skin H_SP138 Spain 2512830 37.7 Unknown MSSP 0 42 1 no Healthy skin H_SP140 Spain 2597272 37.6 Unknown MSSP 1 39 1 no Healthy skin H_SP142 Spain 2597098 37.6 Unknown MSSP 1 39 1 no Healthy skin H_SP143 Spain 2779670 37.6 Unknown MSSP 5 40 1 no sample and MLST. The full data for all isolates can be of 22), even with 100% coverage and identity values for seen in Table S1 of the Supporting information. Pairwise the alleles of the seven genes analysed in the MLST. average nucleotide identity (ANI) values among the 55 Considering all 55 samples together, the average gen- S. pseudintermedius isolates are shown in Table S2. The ome size of S. pseudintermedius was 2.62 Mbp. Compar- average nucleotide identity among the 55 isolates was ing the genomes of the isolates from dogs with pyoderma 99.27%. (n = 33) with those from healthy dogs (n = 22), several In the case of isolates from dogs with pyoderma, ST71 differences were detected. The genomes of the patho- and ST257 were predominant. However, among isolates genic strains were, on average, larger than those of the from healthy dogs, these two STs were absent, and a strains from healthy dogs (2.74 Mbp versus 2.58 Mbp; long list of previously unreported STs was identified (19 P = 4.883e-07; Figure 2a). However, when we analysed © 2021 The Authors. Veterinary Dermatology published by John Wiley & Sons Ltd on behalf of ESVD and ACVD., 32, 654–663. 657 Ferrer et al. a b c d 658 © 2021 The Authors. Veterinary Dermatology published by John Wiley & Sons Ltd on behalf of ESVD and ACVD., 32, 654–663. Whole-genome-sequencing of S. pseudintermedius Figure 2. Meticillin-resistant Staphylococcus pseudintermedius (MRSP) are larger and contain more phages, prophages and antimicrobial resis- tant genes than meticillin-susceptible S. pseudintermedius (MSSP). Box plots show the distribution of (a) genome size, (b) phage and prophage number (c) antimicrobial resistant gene number and (d) virulence factor number of the S. pseudintermedius strains isolated and sequenced in this study. Left column shows results of all S. pseudintermedius (all SP, n = 55), comparing the strains isolated from healthy dogs (turquoise) and those isolated from dogs with pyoderma (blue). Middle column shows results of all S. pseudintermedius (all SP, n = 55), comparing the MSSP strains (turquoise) and the MRSP strains (blue). Right column shows results of the MSSP strains (only MSSP, n = 33), comparing the strains isolated from healthy dogs (turquoise) and those isolated from dogs with pyoderma (blue). Shapiro–Wilk normality tests revealed that the data were not normally distributed. A Wilcoxon rank sum exact test revealed sig- nificant differences in those cases marked by asterisks, denoting significant statistical differences (P < 0.05). the data by comparing the meticillin-resistant S. pseudin- (n = 33) (Table 1), MRSP showed a significant larger gen- termedius (MRSP) isolates, which contained the mecA ome size than the MSSP strains (2.80 Mbp versus gene (n = 22) to the meticillin-susceptible (MSSP) isolates 2.59 Mbp; P = 4.173e-12; Figure 2a). In agreement, COG FUNCTION Comb. Homogeneity Ind. Geo. Homogeneity Ind. Func. Homogeneity Ind. SCG Clusters Max num paralogs Num genes in GC Num contributing genomes H SP143 H SP142 H SP141 H SP140 ACCESSORY H SP138 H SP137 Staphylococcus pseudintermedius all H SP136 H SP135 H SP134 H SP132 H SP127 H SP125 H SP118 H SP097 H SP096 H SP095 H SP094 H SP093 Host H SP082 H SP081 H SP080 H SP079 D SP036 Healthy (22) D SP035 D SP034 D SP032 D SP030 D SP029 Pyoderma (33) D SP028 D SP027 D SP026 D SP025 D SP024 D SP022 D SP021 D SP020 D G099 D G094 D G093 D G089 D G082 Gene clusters D G081 D G078 D G077 D G076 D G072 CORE 2033 D G071 D G067 D G066 D G064 D G063 ACCESSORY 1706 D G062 D G059 D G050 D G040 SINGLETON 479 Singleton gene clusters GC-content Total length SINGLETON CORE ACCESSORY Figure 3. Global pangenome visualization of all Staphylococcus pseudintermedius strains shows 48% of core genome and 52% of accessory genome. Pangenome results of all S. pseudintermedius strains (n = 55), comparing the strains isolated from healthy dogs (turquoise) and the strains iso- lated from dogs with pyoderma (blue). Core genome is by definition the part of the pangenome that is present and shared by all the genomes within the pangenome. Accessory genome is specific for a group of strains within the pangenome and singletons are strain specific genome sequences. Visualization of pangenome analyses carried by ANVI’O. Central dendrogram clustering of samples is ordered by gene cluster presence/ absence. Items order: presence absence (D, Euclidean; L, Ward). © 2021 The Authors. Veterinary Dermatology published by John Wiley & Sons Ltd on behalf of ESVD and ACVD., 32, 654–663. 659 Ferrer et al. a b c d e f 660 © 2021 The Authors. Veterinary Dermatology published by John Wiley & Sons Ltd on behalf of ESVD and ACVD., 32, 654–663. Whole-genome-sequencing of S. pseudintermedius Figure 4. Pangenome analyses show that meticillin-susceptible Staphylococcus pseudintermedius (MSSP) strains are more diverse than methicillin-resistant (MRSP) strains. (a) Split pangenome results of all S. pseudintermedius strains isolated from dogs with pyoderma (blue; n = 33). (b) Split pangenome results of all S. pseudintermedius strains isolated from healthy dogs (turquoise; n = 22). (c) Split pangenome results of all MRSP strains (blue; n = 22). (d) Split pangenome results of all MSSP strains (turquoise; n = 33). (e) Split pangenome results of the MSSP strains (only MSSP, n = 33) isolated from dogs with pyoderma (blue, n = 11). (F) Split pangenome results of the MSSP strains (only MSSP, n = 33) isolated from healthy dogs (turquoise, n = 22). Core genome is by definition the part of the pangenome that is present and shared by all the genomes within the pangenome. Accessory genome is specific for a group of strains within the pangenome and singletons are strain-specific genome sequences. Visualisation of pangenome analyses was carried out using ANVI’O. Central dendrogram clustering of samples is ordered by gene cluster presence/absence. Item order: pres- ence absence (D, Euclidean; L, Ward). MRSP showed, on average, a significantly higher number virulence factor content between pyoderma isolates and of phage and prophages (five versus one; P = 3.924e-06; healthy isolates (Figure 2d), it is noteworthy that four Figure 2b) and antibiotic resistance genes (P = 1.12e-10; genes coding for surface proteins (spsD, spsF, spsP and Figure 2c) than MSSP strains. By contrast, the average spsQ) involved in colonisation by binding to the host’s number of virulence factors was significantly lower in extracellular matrix were present only on pyoderma iso- MRSP than in MSSP isolates (41 versus 42; P = 0.0002; lates. The remaining genes for virulence factors were pre- Figure 2d). Altogether, these results indicate that the dif- sent in some isolates only, without being specific for ferences observed previously between isolates from isolates from healthy dogs or dogs with pyoderma. healthy dogs and dogs with pyoderma (H and D; Figure 2) The current analysis also provided detailed information resulted mainly from the differences between MRSP and about antimicrobial resistance genes (Table S1). Twenty- MSSP strains. Supporting this, when comparing the two of 33 of the isolates from dogs with pyoderma healthy (n = 22) and pyoderma isolates (n = 11; Table 1) (66.6%) carried the mecA gene and were charac- within the MSSP isolates only (n = 33; Table 1), we did terised genotypically as MRSP. However, none of the iso- not observe any differences in either the genome size lates from healthy dogs carried the mecA gene and, (2.58 Mbp versus 2.59 Mbp; P = 0.133; Figure 2a) or the therefore, all were considered MSSP. Further- number of phages and prophages (one versus one; more, mecI and mecR1 genes, two genes involved in P = 0.6195; Figure 2b) or the virulence factors number mecA gene expression, were present in all ST71 isolates, (42 versus 43; P = 0.1491; Figure 2d). However, a signifi- the ST1631 isolate and in one unknown ST. However, the cantly higher number of antibiotic resistance genes was five isolates that belonged to clonal complex 258 (four identified in MSSP isolates from dogs with pyoderma ST258, one ST301) possessed only the mecA gene. when compared to MSSP isolates from healthy dogs Table S1 presents all of the data related to the presence (three versus one; P = 0.03848; Figure 2c). of AMR genes in the 55 strains; those cases where phe- Global pangenome analyses of all the 55 S. pseudinter- notypic sensitivity profiling was available also are indi- medius genomes revealed 48% of core genome and cated. 52% of accessory genome (Figure 3). When pangenome Considering that multidrug resistance (MDR) is defined analyses were carried out comparing the 22 MRSP iso- when the micro-organism is resistant to at least three lates against the 33 MSSP isolates, we observed a larger families of antibiotics, we observed an MDR gene profile core genome within the MRSP than within the MSSP in 81.82% (27 of 33) of the S. pseudintermedius isolates pangenome (63% versus 52%; Figure 4c, d). These from lesional skin versus 9.09% (two of 22) of the iso- results suggest that the MSSP pangenome is more lates from healthy dogs. Among the 27 MDR strains iso- diverse than the MRSP pangenome. In agreement with lated from dogs with pyoderma, 22 held the mecA gene this hypothesis, the specific singleton genome within the (Table 1). MSSP pangenome was larger than in the MRSP pangen- ome (17% versus 9%; Figure 4c, d). Supporting these Discussion results, 70% of the MRSP isolates were identified as ST71 MLST, while the MSSP comprised a mixture of dif- In the present study, we were able to sequence and de ferent MLSTs, most as yet unidentified and thus classi- novo assemble the complete genome of multiple isolates fied as unknown (Table S1). Comparison of the pyoderma of S. pseudintermedius using Nanopore technology. The and healthy cases within the MSSP isolates only, showed method is notably faster, simpler and cheaper than other similar core and accessory genome sizes (60% versus sequencing systems, such as the traditional Illumina sys- 64%; Figure 4e, f). Altogether, the pangenome analyses tem. In addition, the use of ANVI’O allowed for easy com- indicate that differences detected between isolates from parison of the genomes. healthy dogs and from dogs with pyoderma are mainly The complete genome analysis of S. pseudintermedius the result of differences between MRSP and MSSP is undoubtedly a very useful tool to characterise the dif- strains. ferent isolates. In particular, it allowed the identification Fifty virulence genes were identified, including acces- and location of virulence and antimicrobial resistance fac- sory gene regulators (agr A, B, C, D), adhesins and biofilm tor genes as well as the presence of phages, and is very formation genes (sps A-H,ICA A-D,ebps), toxins (expA, likely to help to better understand the pathogenesis of expB, siet, speta) and invasins (Luk F, S; Hlb, Coa). Thirty- canine pyodermas. two (58.18%) of the virulence factors were present in all The S. pseudintermedius isolates from our study have 55 isolates. Even though there are no differences in an average genome size of 2.62 Mbp. This size is similar © 2021 The Authors. Veterinary Dermatology published by John Wiley & Sons Ltd on behalf of ESVD and ACVD., 32, 654–663. 661 Ferrer et al. to that reported by other authors29-31 and the NCBI which seems to go against the paradigm constructed S. pseudintermedius reference genome (NC_014925.1). from the study by Pinchberk and colleagues stating that This is an example of an open genome, with a core gen- the S. pseudintermedius causing skin infections very ome of 48% and an accessory genome of 52%. likely originate from commensal S. pseudintermedius The S. pseudintermedius isolates from pyoderma populations.5 The difference also could be a conse- cases had a significantly larger genome than isolates from quence of the fact that all ST71 and ST258 strains iso- healthy skin. However, when further comparative analy- lated so far (https://pubmlst.org) are MRSP and in the ses were performed, it was found that the size difference group of healthy dogs all isolates were MSSP. In any was to the result of the greater presence of MRSP in the case, our results should be interpreted with caution group of isolates from dogs with pyoderma. This larger because the number of samples was small, especially genome size in MRSP compared to MSSP (2.80 Mbp ver- from healthy dogs. Extensive research and sequencing sus 2.59 Mbp) was undoubtedly a consequence of the of S. pseudintermedius from healthy dogs are needed increased presence of phages, prophages and antimicro- to clarify whether there are typically pathogenic STs and bial resistance factors in MRSP. others that are associated only with commensal beha- Interestingly, hardly any differences were found in the viour. Additionally, to identify genetic changes associ- presence of virulence factor genes among the different ated with the transformation of a commensal micro- strains. Only a few genes encoding surface proteins (sps organism into a pathogen we should study isolates BF) were found, exclusively or much more frequently in obtained from the same dog, from both lesional and samples from pyoderma cases. The surface proteins of nonlesional areas. S. pseudintermedius express binding activity to compo- In summary, our approach has allowed the sequencing nents of the host’s extracellular matrix (ECM), including and de novo assembly of the complete genome of 55 fibronectin, fibrinogen and cytokeratin10, and they are S. pseudintermedius strains isolated from healthy dogs considered to be virulence factors associated with bacte- and from dogs with pyoderma, and their analysis and rial survival, immune evasion and biofilm formation.32,33 comparison. This technique allows much more precise Our results support previous data indicating that these characterisation of S. pseudintermedius populations in proteins may play a role in staphylococcal colonisation dog skin than techniques usedpreviously , such as PFGE. and/or infection.33-35 Interestingly, however, most viru- Our hope is that this new approach will lead to a better lence factors are present also in strains isolated from understanding of the pathogenesis of canine pyodermas. healthy dogs, indicating that commensal staphylococcal populations already have full pathogenic potential. In any References case, it is worth noting that this genomic approach inves- tigates only the presence of certain genetic elements in 1. Paul NC, B€ argman SC, Moodley A, et al. Staphylococcus pseud- the bacterial genome and not their expression or function- intermedius colonization patterns and strain diversity in healthy ality. In the case of both antimicrobial resistance and viru- dogs: a cross-sectional and longitudinal study. Vet Microbiol 2012; 160: 420–427. lence factors, it is necessary to complement genetic 2. Rubin JE, Chirino-Trejo M. Prevalence, sites of colonization, studies with functionality tests. anantimicrobial resistance among Staphylococcus pseudinter- The isolates from lesional skin of dogs with pyoderma medius isolated from healthy dogs in Saskatoon, Canada. J Vet had a much higher number of antimicrobial resistance Diagn Invest 2011; 23: 351–354. genes than isolates from skin of healthy dogs (Figure 2c), 3. Bannoehr J, Guardabassi L. Staphylococcus pseudintermedius an expected finding that probably reflects prior antibiotic in the dog: taxonomy, diagnostics, ecology, epidemiology and exposure. This finding, again, resulted from the presence pathogenicity. Vet Dermatol 2012; 23: 253-66, e51-2. 4. Loeffler A, Lloyd DH. What has changed in canine pyoderma? A of MRSP in this group (22 of 33 isolates were MRSP). Fig- narrative review. Vet J 2018; 235: 73–82. ure 2c shows the greater presence of AMRG in MRSPs 5. Pinchbeck LR, Cole LK, Hillier A, et al. Genotypic relatedness of compared to MSSPs. staphylococcal strains isolated from pustules and carriage sites The present approach allowed not only the detection of in dogs with superficial bacterial folliculitis. Am J Vet Res 2006; antimicrobial resistance genes, but also its location in the 67: 1,337–1,346. genome. For instance, three ST71 isolates and one ST258 6. Jain M, Olsen HE, Paten B, et al. The Oxford Nanopore MinION: delivery of nanopore sequencing to the genomics community. isolate harboured the tetK gene in another contig rather Genome Biol 2016; 17: 239. than the chromosome, which was probably a plasmid, as ~es J, Cusco 7. Vin  A, Francino O. Hybrid assembly from a patho- these isolates harboured the replicon rep7_1_repC too genic methicillin- and multidrug-resistant Staphylococcus pseud- (Table S1). Three ST258 and one ST301 isolates har- intermedius strain Isolated from a case of canine otitis in Spain. boured the tetM gene in the chromosome (Table S1). Microbiol Resour Announc 2020; 9: e01121–19. Five ST71 and two unknown ST isolates harboured the 8. Gardy JL, Loman NJ. Towards a genomics-informed, real-time, chloramphenicol resistance gene in the same contig as global pathogen surveillance system. Nat Rev Genet 2018; 19: 9–20. the replicon rep7_7_rep(pKH7), suggesting that this gene 9. Lu H, Giordano F, Ning Z. Oxford nanopore MinION sequencing is linked to a plasmid in those isolates (Table S1). and genome assembly. Genomics Proteomics Bioinformatics An unexpected result was the difference in the 2016; 14: 265–279. MLSTs identified in animals with pyoderma and in 10. Hillier A, Lloyd DH, Weese JS, et al. Guidelines for the diagnosis healthy animals. In the strains from dogs with pyo- and antimicrobial therapy of canine superficial bacterial folliculitis derma, the STs identified were those most frequently (Antimicrobial Guidelines Working Group of the International reported in Europe (ST71, ST258). However, none of Society for Companion Animal Infectious Diseases). Vet Derma- tol 2014; 25: 163-e43. these STs were detected in samples from healthy dogs, 662 © 2021 The Authors. Veterinary Dermatology published by John Wiley & Sons Ltd on behalf of ESVD and ACVD., 32, 654–663. Whole-genome-sequencing of S. pseudintermedius 11. De Coster W, D’Hert S, Schultz DT, et al. NanoPack: visualizing isolates from healthy dogs and dogs with pyoderma. Microbiol and processing long-read sequencing data. Bioinformatics 2018; Res Announc 2021; 10: e00152–21. 34: 2666–2669. 29. Tse H, Tsoi HW, Leung SP, et al. Complete genome sequence 12. Juul S, Izquierdo F, Hurst A, et al. What’s in my pot? Real-time of the veterinary pathogen Staphylococcus pseudintermedius species identification on the MinIONTM. BioRxiv 2015. https:// strain HKU10-03, isolated in a case of canine pyoderma. J Bacte- doi.org/10.1101/030742. riol 2011; 193: 1,783–1,784. 13. Kolmogorov M, Yuan J, Lin YU, et al. Assembly of long, error-prone 30. Eichhorn I, Lu€bke-Becker A, Lapschies A-M, et al. Draft genome reads using repeat graphs. Nat Biotechnol 2019; 37: 540–549. sequence of Staphylococcus pseudintermedius strain 13– 14. Li H. Minimap2: Pairwise alignment for nucleotide sequences. 13613, isolated from a case of canine pyoderma. Microbiol Bioinformatics 2018; 34: 3,094–3,100. Resour Announc 2020; 9: e00027–20. 15. Parks DH, Imelfort M, Skennerton CT, et al. CheckM: Assessing 31. Cao W, Hicks K, White A, et al. Draft genome assemblies of two the quality of microbial genomes recovered from isolates, single Staphylococcus pseudintermedius strains isolated from canine cells, and metagenomes. Genome Res 2015; 25: 1,043–1,055. skin biopsy specimens. Microbiol Resour Announc 2020; 9: 16. Hunt M, Silva ND, Otto TD, et al. Circlator: Automated circular- e00369–20. ization of genome assemblies using long sequencing reads. 32. Foster TJ. Colonization and infection of the human host by Genome Biol 2015; 16: 294. staphylococci: adhesion, survival and immune evasion. Vet Der- 17. Tatusova T, DiCuccio M, Badretdin A, et al. NCBI prokaryotic gen- matol 2009; 20: 456–470. ome annotation pipeline. Nucleic Acids Res 2016; 44: 6,614–6,624. 33. Sung JM-L, Lloyd DH, Lindsay JA. Staphylococcus aureus host 18. Jain C, Rodriguez-R LM, Phillippy AM, et al. High throughput specificity: comparative genomics of human versus animal iso- ANI analysis of 90K prokaryotic genomes reveals clear species lates by multi-strain microarray. Microbiology 2008; 154: 1,949– boundaries. Nat Commun 2018; 9: 5,114. 1,959. 19. Eren AM, Esen OC,€ Quince C, et al. Anvi’o: An advanced analysis 34. Bannoehr J, Ben Zakour NL, Reglinski M, et al. Genomic and and visualization platform for ’omics data. PeerJ 2015; 3: e1319. surface proteomic analysis of the canine pathogen Staphylococ- 20. Davis JJ, Wattam AR, Aziz RK, et al. The PATRIC Bioinformatics cus pseudintermedius reveals proteins that mediate adherence Resource Center: expanding data and analysis capabilities. to the extracellular matrix. Infect Immun 2011; 79: 3074–3086. Nucleic Acids Res 2020; 48(D1), D606–D612. 35. Phumthanakorn N, Chanchaithong P, Prapasarakul N. Develop- 21. Stamatakis A. RAxML version 8: a tool for phylogenetic analysis ment of a set of multiplex PCRs for detection of genes encoding and post-analysis of large phylogenies. Bioinformatics 2014; 30: cell wall-associated proteins in Staphylococcus pseudinter- 1312–1313. medius isolates from dogs, humans and the environment. J 22. Larsen MV, Cosentino S, Rasmussen S, et al. Multilocus Microbiol Methods 2017; 142: 90–95. sequence typing of total-genome-sequenced bacteria. J Clin Microbiol 2012; 50: 1,355–1,361. 23. Seemann T. ABRicate. GitHub 2017. Available at: https://github. Supporting Information com/tseemann/abricate Accessed Oct 6, 2021. 24. Jia B, Raphenya AR, Alcock B, et al. CARD 2017: Expansion and Additional Supporting Information may be found in the model-centric curation of the comprehensive antibiotic resis- online version of this article. tance database. Nucleic Acids Res 2017; 45(D1): D566–D573. Table S1. Data for the 55 sequenced genomes. For each 25. Carattoli A, Zankari E, Garcıa-Fernandez A, et al. In silico detec- genome the following parameters are indicated: MLST tion and typing of plasmids using plasmidfinder and plasmid mul- (multilocus site typing); location (country of origin of sam- tilocus sequence typing. Antimicrob Agents Chemother 2014; ple); genome size (bp); number of contigs; longest contig 58: 3895–3903. (bp); number of phages and prophages detected; pre- 26. Starikova EV, Tikhonova PO, Prianichnikov NA, et al. Phigaro: high-throughput prophage sequence annotation. Bioinformatics sence of different antimicrobial resistance genes and pre- 2020; 36: 3,882–3,884. sence of plasmids. 27. Roux S, Enault F, Hurwitz BL, et al. VirSorter: Mining viral signal Table S2. Pairwise average nucleotide identity (ANI) from microbial genomic data. PeerJ 2015; 3: e985. values among the 55 S. pseudintermedius isolates. 28. Francino O, Pe rez D, Vin ~es J, et al. Whole-genome sequencing Values >99.995 are indicated in bold. and de novo assembly of 61 Staphylococcus pseudintermedius © 2021 The Authors. Veterinary Dermatology published by John Wiley & Sons Ltd on behalf of ESVD and ACVD., 32, 654–663. 663