RESEARCH ARTICLE Revealing Genomic Insights of the Unexplored Porcine Pathogen Actinobacillus pleuropneumoniae Using Whole Genome Sequencing Judith Guitart-Matas,a,b,c Narjol Gonzalez-Escalona,d Meghan Maguire,e Anna Vilaró,e Jaime Martinez-Urtaza,f Lorenzo Fraile,g Lourdes Migura-Garciaa,b,c a Joint Research Unit IRTA-UAB in Animal Health, Animal Health Research Centre (CReSA), Campus of the Autonomous University of Barcelona (UAB), Bellaterra, Catalonia, Spain b IRTA, Animal Health Program, Animal Health Research Centre (CReSA), Campus of the Autonomous University of Barcelona (UAB), Bellaterra, Catalonia, Spain c OIE Collaborating Centre for the Research and Control of Emerging and Re-Emerging Swine Diseases in Europe (IRTA-CReSA), Bellaterra, Barcelona, Spain d Food and Drug Administration, Silver Spring, Maryland, United States e Pig Sanitation Group, Lleida, Spain Department of Genetics, Autonomous University of Barcelona, Bellaterra, Spain f g Department of Animal Science, ETSEA, University of Lleida, Lleida, Spain ABSTRACT Actinobacillus pleuropneumoniae (APP) is the causative agent of pleuro- pneumonia in pigs, one of the most relevant bacterial respiratory diseases in the swine industry. To date, 19 serotypes have been described based on capsular polysaccha- ride typing with significant virulence dissimilarities. In this study, 16 APP isolates from Spanish origin were selected to perform antimicrobial susceptibility tests and compara- tive genomic analysis using whole genome sequencing (WGS). To obtain a more compre- hensive worldwide molecular epidemiologic analyses, all APP whole genome assemblies available at the National Center for Biotechnology Information (NCBI) at the time of the study were also included. An in-house in silico PCR approach enabled the correct serotyp- ing of unserotyped or incorrectly serotyped isolates and allowed for the discrimination between serotypes 9 and 11. A pangenome analysis identified the presence or absence of gene clusters to be serotype specific, as well as virulence profile analyses targeting the apx operons. Antimicrobial resistance genes were correlated to the presence of spe- cific plasmids. Altogether, this study provides new insights into the genetic variability within APP serotypes, correlates phenotypic tests with bioinformatic analyses and mani- fests the benefits of populated databases for a better assessment of diversity and vari- ability of relatively unknown pathogens. Overall, genomic comparative analysis enhances the understanding of transmission and epidemiological patterns of this species and sug- gests vertical transmission of the pathogen, including the resistance genes, within the Spanish integrated systems. IMPORTANCE Pleuropneumonia is one of the most relevant respiratory infections in the swine industry. Despite Actinobacillus pleuropneumoniae (APP) being one of the Editor Artem S. Rogovskyy, Texas A&M University most important pathogens in the pig production, this is the first comparative study Copyright © 2022 Guitart-Matas et al. This is including all available whole genome sequencing data from NCBI. Moreover, this an open-access article distributed under the study also includes 16 APP isolates of Spanish origin with known epidemiological terms of the Creative Commons Attribution 4.0 International license. relationships through vertical integrated systems. Genomic comparisons provided a Address correspondence to Lourdes Migura- deeper understanding of molecular and epidemiological knowledge between differ- Garcia, lourdes.migura@irta.cat. ent APP serotypes. Furthermore, determination of resistance and toxin profiles The authors declare no conflict of interest. allowed correlation with the presence of mobile genetic elements and specific sero- Received 4 April 2022 type, respectively. Accepted 23 June 2022 Published 20 July 2022 KEYWORDS Actinobacillus, whole genome sequencing, swine July/August 2022 Volume 10 Issue 4 10.1128/spectrum.01185-22 1 Genomics of Actinobacillus pleuropneumoniae in Pigs Microbiology Spectrum A ctinobacillus pleuropneumoniae (APP) is the pathogen responsible of one of the most important bacterial respiratory diseases in the swine industry worldwide (1). Thus, it is listed among the 10 most important pathogens affecting pig production (2). In most countries, this disease is not under official control, and therefore, exact and updated information about its prevalence is difficult to know since reporting cases is not mandated by law. Animals infected with APP can present a range of clinical symp- toms that vary from acute to chronic and even subclinical, but when outbreaks occur in field conditions, a sudden increase in mortality is usually observed (1). Overall, this disease causes large economic losses to the pig industry due to increased mortality rates, reduced growth rates, and cost of the control measures such as antibiotic treat- ments and vaccines. This pathogen presents a huge variability worldwide. Currently, 19 serotypes have been described for this pathogen that differ significantly in virulence (1, 3). Different factors such as management practices (weaning age for example), time of the year, and production systems are also risk factors associated with the emergence of this pathogen in each particular farm or pig integration company (4). Interestingly, 80% of the swine industry in Spain is integrated, with a hierarchical pyramid structure. Generally, these pyramids integrate vertically the different levels of the production within the same system but separating physically the different phases of the rearing cycle. For example, one company holds the genetic selection of breeders (grand- mothers) in a farm with high-health status that supplies the gilts for multiplication farms. When these gilts are approximately 6 months of age, they are moved to the different multiplication farms (mothers) where they will be mated to provide piglets. After weaning, piglets of approximately 30 kg are transported to finishing farms, where they remain until slaughter for human consumption. These vertically inte- grated systems may explain the circulation of pathogens and resistance traits from the top of the pyramid (breeders) to the bottom (piglets). Furthermore, one can hypothesize that knowledge of the epidemiology of the APP circulating in the pro- duction pyramid may help to implement preventive medicine programs for an effi- cient control of the disease in the whole production system. Moreover, the European legislation on veterinary medicinal products recommends using epidemiological in- formation as a sound criterion to select the most suitable antimicrobial to be used in each clinical case (5). Thus, it has been recently proposed to determine the MIC for a battery of antimicrobials in one clinical case and use these epidemiological data in future clinical cases if the sow origin is the same (6). However, this approach has not been validated from a scientific point of view. Alternatively, the characterization of the APP isolates circulating in these integrated systems is essential to apply control measures, since commercial vaccines are usually based on bacterins that provide limited cross-protection between serotypes (7). Currently, the most common typing scheme for APP is based on the antigenic properties of the capsule polysaccharides, with 19 serotypes described up to date (3, 8). Virulence mechanisms conferred by the repeat-in-toxin (RTX) family (toxins Apx I to IV) are involved in the development of disease, providing different cytotoxic effects depending on the toxins produced by each particular APP. Additionally, serotype-specific secretion of Apx toxins is commonly reported (9–11). Based on these typing methods, several studies have attempted to describe the epidemiology of the disease in different coun- tries (12, 13). Recently, whole genome sequencing (WGS) has also been performed to unravel the genomic variability of APP, but mainly for serotype 8, and a scarce number of APP isolates have been sequenced over the last few years (14, 15). To our knowledge, there is a lack of studies applying WGS for epidemiological studies of this disease, and there are few APP DNA sequences published in public databases, especially of Spanish origin. Therefore, the first aim of this study is to apply comparative genomics to assess virulence, resistance, and phylogeny of all available isolate assemblies in public data- bases. Furthermore, Spanish APP lineages in related farms are epidemiologically ana- lyzed to determine the transmission of this respiratory pathogen within these integrated July/August 2022 Volume 10 Issue 4 10.1128/spectrum.01185-22 2 Genomics of Actinobacillus pleuropneumoniae in Pigs Microbiology Spectrum FIG 1 Epidemiological link between fattening, multiplication or mothers' farm (M), and selection or grandmothers' farm (GM) for the 16 APP isolates selected for whole genome sequencing analysis. GM2 and GM4 (*) farms shared the same origin of sows during foundation of both farms. systems. Finally, WGS has been used to demonstrate the suitability of the epidemiologi- cal approach to support prudent use of antimicrobials for this bacterium in swine medicine. RESULTS General genomic features of selected isolates. Whole genome sequencing data from the 16 APP isolates belonging to this study (Fig. 1) allowed the generation of ge- nome assemblies with high quality. A total set of 31 additional APP assemblies avail- able in NCBI at the time of the study (April 2021) was downloaded and checked for quality. Only one sample was discarded because its genome length was four times lon- ger compared to the rest of the genomes. The final list of 46 isolates included in the analysis is detailed in Table 1. The genome of APP strain S4074 available at NCBI (GenBank accession number CP030753.1) with a genome size of 2,318,257 bp and GC content of 41.24%, was used as the representative and reference for that species for the analysis. The results obtained from the QUAST analysis of all the assemblies determined that the average genome size for this species was 2.3 Mbp and the GC content ranged from 41.16% to 41.20%. Serotyping using in-house in silico PCRs. The serotype of four APP of the data set downloaded from NCBI was not available or was not included in the description. The assembly IDs of these isolates were: 42650_C01 (strain NCTC11384), 42650_D01 (strain NCTC11407), 57675_E01 (strain NCTC10976), and ASM1676271v1 (strain 1140). In silico PCR allowed the accurate serotyping of these isolates that were identified as serotypes 4, 6, 2, and 1, respectively. Moreover, two isolates were found to be incorrectly linked to a specific reference strain and corresponding serotype. The genome assemblies ASM16709v1 and ASM17865v1, linked to strains 4074 and N273, respectively, were cor- rectly identified as serotypes 5 and 7. Separately, the alignment of the total 15 isolates belonging to serotypes 9 and 11 showed a base deletion at the end of the cps gene (cpsF) in all of them, except for the known serotype 9 isolate (strain CVJ13261). Moreover, the length of the PCR product for these isolates was 1,242 bp, confirming Spanish isolates from M2 to M5 as serotype 11. Phylogenetic reconstruction and in silico determination of virulence and anti- microbial resistance profiles. To infer the phylogenetic relationships between the APP isolates, the core genome alignment based on SNPs was used. A total of 8,028 sites were July/August 2022 Volume 10 Issue 4 10.1128/spectrum.01185-22 3 Genomics of Actinobacillus pleuropneumoniae in Pigs Microbiology Spectrum TABLE 1 List of isolates selected for the analysis (n = 46) and metadata information including serotype, country of isolation, and year of isolationa This study Country of Selection Multiplication ID Serotype isolation Yr of isolation farm (GM) farm (M) Biosample SRA accession WGS accession UP1971607 13 Spain 2019 1 1 SAMN23242902 SRR16969607 JAJMUP000000000 UP1971182 13 Spain 2019 1 1 SAMN23242903 SRR16969606 JAJMUO000000000 UG1871511 13 Spain 2018 1 1 SAMN23242904 SRR16969605 JAJMUN000000000 UG1970106 11 Spain 2019 2 2 SAMN23242905 SRR16969618 JAJMUM000000000 UG1970008 11 Spain 2019 2 2 SAMN23242897 SRR16969612 JAJMUU000000000 UG1970316 11 Spain 2019 2 3 SAMN23242906 SRR16969617 JAJMUL000000000 UG1871964_S21 11 Spain 2018 2 3 SAMN23242907 SRR16969616 JAJMUK000000000 UG1871964_S6 11 Spain 2018 2 3 SAMN23242898 SRR16969611 JAJMUT000000000 UG1970170 11 Spain 2019 2 3 SAMN23242901 SRR16969608 JAJMUQ000000000 UG1971013 11 Spain 2019 3 4 SAMN23242896 SRR16969619 JAJMUV000000000 UP1971676 11 Spain 2019 3 4 SAMN23242908 SRR16969615 JAJMUJ000000000 UG1971163 11 Spain 2019 4 5 SAMN23242909 SRR16969614 JAJMUI000000000 UG1970277 11 Spain 2019 4 5 SAMN23242900 SRR16969609 JAJMUR000000000 UG1970269 11 Spain 2019 4 5 SAMN23242899 SRR16969610 JAJMUS000000000 UG1970987 11 Spain 2019 4 5 SAMN23242910 SRR16969613 JAJMUH000000000 UG1970667 11 Spain 2019 4 5 SAMN23242895 SRR16969620 JAJMUW000000000 NCBI assembly database Country of GenBank ID Serotype isolation Yr of isolation Strain accession Reference 42650_C01 4 USA 1980 NCTC11384 LS483358.1 Kilian, M. et al. (16) 42650_D01 6 Denmark 1971 NCTC11407 Pohl, S. et al. (17) 56750_E01 2 Denmark 1973 NCTC10976 LR134515.1 ASM1588v1 5 2007 L20 CP000569.1 Foote, S. et al. (18) ASM1668v1 3 China 2008 JL03 CP000687.1 Xu, Z. et al. (19) ASM16709v1 5 Argentina 2003 4074 Xu, Z. et al. (20) ASM1676271v1 1 China 2017 1140 No reference ASM1735746v1 5 China 2021 App6 CP026009.1 No reference ASM17849v2 1 Argentina 2018 4074 CP029003.1 Xu, Z. et al. (20) ASM17851v1 2 2010 S1536 ASM17853v1 4 Australia 2010 M62 ASM17855v1 6 Australia 2010 Femo ASM17857v1 9 Netherlands 2010 CVJ13261 ASM17859v1 10 2010 D13039 ASM17861v1 11 Netherlands 2010 56153 ASM17863v1 12 Australia 2010 1096 ASM17865v1 7 Hungary 2010 N273 Xu, Z. et al. (20) ASM17927v1 2 Australia 2010 4226 Zhan, B. et al. (21) ASM17929v1 6 Australia 2010 Femo ASM2040v1 7 Canada 2008 AP76 CP001091.1 Xu, Z. et al. (20) ASM29591v1 7 China 2012 S8 Li, G. et al. (22) ASM329038v1 1 Argentina 2018 S4074 CP030753.1 Dona, V. and Perreten, V. (23) ASM343140v1 1 South Korea 2017 16:00 CP022715.1 No reference ASM81744v1 8 Brazil 2011 1022 Pereira, M. et al. (24) ASM81746v1 8 Brazil 2007 460 ASM81748v1 8 Brazil 2007 518 ASM81751v1 8 Brazil 2009 780 ASM81752v1 8 Brazil 2006 5651 ASM81753v1 8 Brazil 2008 597 Pereira, M. et al. (24) MIDG2331 8 UK 2015 MIDG2331 LN908249.1 Bossé, J.T. et al. (25) aA total of 16 isolates derived from this study and metadata also includes both selection and multiplication origin farms. Separately, 30 isolates were selected from the NCBI database. used to construct the phylogenetic tree depicted in Fig. 2. Despite the low number of available isolates per serotype, distinct clades were clearly grouped by serotype. As previ- ously reported, the virulent isolates of serotypes 1, 9, and 11 were aggregated in a com- mon clade. Matrix counts of dissimilarity identified a maximum of 140 counts between isolates belonging to these serotypes. Longer distances were observed between this clade and the rest of serotypes, all of them above 2,000 counts of dissimilarity. The follow- ing serotype group with highest dissimilarity was serotype 5, with 183 maximum counts, followed by serotypes 7 and 8 with a maximum of 39 counts per cluster (Table S1). Within Spanish isolates, breeds 1 and 2 were clearly split into two different clusters. Isolates collected from finishing farms supplied by M1 were considered identical. All iso- lates belonging to breed 2 were serotype 11 and were detected in farms supplied by M2 to M5. Another observation within this cluster was the presence of the Dutch isolate among Spanish isolates from multiplication farms M3 and M5 (Table S1). This geographical dispersion was also observed in other clusters, such as the relevant similarity between serotype 6 isolates from Australia and the Danish strain. This similarity also occurred within serotype 2, and within serotypes 7 and 4, from Canada and USA, respectively. July/August 2022 Volume 10 Issue 4 10.1128/spectrum.01185-22 4 Genomics of Actinobacillus pleuropneumoniae in Pigs Microbiology Spectrum FIG 2 Virulence and resistance profile of the 46 APP isolates in a phylogenetic reconstruction. Columns from left to right indicate, multiplication farm (M), selection farm (GM), virulence profile including apxI, apxII and apxIII genes, plasmid and phage presence, and resistance profile. Phylogenetic reconstruction based on SNPs is represented on the left panel. Branch lengths and bootstrap support values are represented on the top and bottom of branches, respectively. Origin legend is specified on the top left of the figure. Serotype of each isolate is specified in the term before the country origin information of the isolate IDs. Recombination tools did not detect recombination accurately, as putative events were masked with regions with high density of base substitutions. This observation suggested that the different serotypes are remarkably diverse, and the data set should be split into closer related serotypes to appropriately assess recombination. Moreover, root-to-tip regression analyses suggested weak temporal signal, insufficient to proceed with evolutionary analyses. The previously reported correlation between specific serotypes and the presence of Apx toxin genes was confirmed for apxI, apxII, and apxIII genes, as represented in Fig. 2. For these annotated genes, the identity was higher than 98% and the coverage was 100%. The in silico analysis for the presence of the apxIV gene, which codes for the ApxIV toxin, known to be secreted by all serotypes, produced inconsistent results regarding coverage percentages. Hence, as repeat units are known to lead to assembly errors, these features were further studied with the UGENE repeat analysis tool. Results showed that the apxIV gene contains many repeat regions that made its de novo as- sembly difficult. Fig. 3 depicts these regions in a 1,759 bp window with several repeat units from positions 3,388 to 5,146 of the apxIV gene. In addition, analysis with the Virulence Factor DataBase (VFDB) identified only one virulence factor present in all 46 A. pleuropneumoniae isolates. This virulence factor is the phosphoheptose isomerase (gmhA/lpcA). The in silico antimicrobial resistance profiles analysis for the 46 APP, detected using the CARD database, identified the presence of 13 resistance genes (Fig. 2). A total of 18 isolates harbored at least one gene conferring antimicrobial resistance to different anti- microbials correlated to the presence of plasmids (Table 2). Interestingly, a correlation was observed between isolates from serotypes 13 and 7, known to have the same Apx profile, and a similar resistance profile. Isolates from serotype 8 presented more vari- ability, but all of them harbored at least one resistance gene. Two of the three serotype 5 isolates did not yield resistant genes, while the most recent isolate from 2021 (ASM1735746v1), carried a plasmid and harbored 8 resistance genes. July/August 2022 Volume 10 Issue 4 10.1128/spectrum.01185-22 5 Genomics of Actinobacillus pleuropneumoniae in Pigs Microbiology Spectrum FIG 3 Repeat units detected by UGENE in nucleotide sequence of apxIV CDS from GenBank accession number AF021919.1. Each repeat unit is represented with a turquoise arrow and matches are linked with black lines. All repeat units share 97% of identity or higher. Regarding the Spanish isolates, sequencing data showing the presence of resist- ance genes agreed with the phenotypic results obtained by minimal inhibitory concen- tration (MIC). Three distinct phenotypes were identified (Table S2): i) resistance to amoxicillin, doxycycline, enrofloxacin, marbofloxacin, and oxytetracycline in all isolates of serotype 13 supplied by multiplication farm M1, harboring plasmid pB1005 associ- ated to Pasteurella multocida (Table 2); ii) a pan-susceptible profile corresponding to 11 isolates of serotype 11; and iii) the remaining two isolates of serotype 11 (UG1970106 and UG1970008) that exhibited resistance to sulfamethoxazole/trimethoprim, doxycy- cline, and oxytetracycline. These two isolates were supplied by multiplication farm M2 where a tetracycline treatment was employed and harbored the plasmid p11745. A total of 21 potential active prophages were detected among the 46 APP isolates with the Prophage Hunter tool (Table S3). The most common prophage identified was the Mannheimia phage (n = 18). The remaining three prophages identified were Lactobacillus, Streptococcus, and Stenotrophomonas phages. All of them had a score above 0.84. Artemis visualization of genomic islands did not show a pathogenicity island pat- tern among isolates or serotypes. Between 20 and 32 genomic islands were predicted by the Alien Hunter software for each isolate’s genome and the lengths of these islands were also diverse (data not shown). Pangenome analysis. Pangenome analysis with anvi’o identified a total of 3,200 gene clusters and 102,596 gene calls. The core and soft-core bins, including gene clusters shared by more than the 99% or between 95 to 99% of the isolates, respectively, con- tained 1,826 and 80 gene clusters, respectively. In addition, the shell bin, comprising gene clusters present in 15 to 95% of the isolates, contained 461 gene clusters, while the cloud bin, representing the gene clusters present in less than the 15% of the isolates, contained 833 gene clusters. TABLE 2 Plasmids identified with PLSDB with a minimum identity of 0.98 and resistance genes known to be harbored in those plasmidsa Length Resistance Isolate ID NCBI RefSeq (bp) GC (%) Taxon Plasmid name genes UG1871511_13_Spain_2018 NC_012215.1 4237 48.12 Pasteurella multocida pB1005b sul2 UP1971182_13_Spain_2019 strA UP1971607_13_Spain_2019 ASM343140_1_SouthKorea_2017 NZ_CP022716.1 7699 60.93 APP unnamed1 floR ASM1676271_1_China_2017 NZ_MT230378.1 1179 42.83 E. coli pESBL87 tet(B) UG1970008_11_Spain_2019 NC_013546.1 5486 40.09 APP p11745 tet(B) UG1970106_11_Spain_2019 ASM1735746_5_China_2021 NZ_KX434882.1 4848 58.75 Klebsiella pneumoniae pKP2442_7c331 floR ASM2040_7_Canada_2008 NC_010942.1 5685 41.5 APP APP7_Ac rob-1 ASM17865_7_Hungary_2010 NC_019176.1 4613 41.45 Haemophilus influenzae pB1000c rob-1 ASM29591_7_China_2012 NC_007098.1 3156 46.51 APP pKMA2425 sul2 MIDG2331_8_UK_2015 NZ_MT230378.1 1179 42.83 E. coli pESBL87 tet(B) ASM81744_8_Brazil_2011 NZ_MH457196.1 5128 35.61 APP p780 tet(B) ASM81746_8_Brazil_2007 ASM81748_8_Brazil_2009 NZ_KT355773.1 3937 52.78 APP p518 floR aph(30)-Ib ASM81751_8_Brazil_2006 NZ_MH457196.1 5128 35.61 APP p780 tet(B) ASM81752_8_Brazil_2008 NC_009625.1 4065 45.41 APP pARD3079 sul2 ASM81753_8_Brazil_2007 aOnly isolates identified with a plasmid are listed. Results from plasmidSPAdes were in agreement with PLSDB. bResistance region was 100% identical to the ABB7_B plasmid locus of APP. cAlignment of both plasmid sequences shared 81% coverage with 99.8% of identity. July/August 2022 Volume 10 Issue 4 10.1128/spectrum.01185-22 6 Genomics of Actinobacillus pleuropneumoniae in Pigs Microbiology Spectrum For proper identification of specific gene calls and clusters of orthologous gene (COG) functions, the full pangenome was split into two independent pangenomes: the core, including the gene clusters of the core and the soft-core, and the accessory, including the shell and the cloud gene clusters. Fig. 4A represents the full pangenome, while Fig. 4B only includes core and soft-core gene clusters. Shell and singleton gene clusters, representing the accessory, are shown in Fig. 4C. Selected APP isolates were ordered by presence/absence of gene clusters and col- ored by serotype. As observed in Fig. 4C, a matrix representing ANI values correlates with distribution of presence/absence of gene clusters and serotype identification. Moreover, the previously observed clustering and cross-reactivity of serotypes 1, 9, and 11 is also detected in this matrix. This block is clearly clustered apart from the other serotypes, and includes, from right to left, the four isolates of serotype 1, the 13 iso- lates of this study identified as serotype 11, and the two isolates from The Netherlands, belonging to serotypes 9 and 11. From the accessory bin, which included the gene clusters represented in less than 95% of the isolates, specific gene calls were identified, and COG functions putatively related to resistance and virulence were annotated. Within the accessory bin, the shell (gene clusters presented in 15% to 95% of the genomes) integrated several gene calls related to defense mechanisms and virulence factors, such as proteins VagC or RhuM, and proteins related to RTX toxins. Also, plasmid-related proteins were identified, such as proteins ParE, HigB, and VapI, transposon-related proteins, the SOS response repressor LexA, and multidrug resistance (MDR) efflux pumps. The cloud (gene clusters presented in less than 15% of the genomes) included gene calls associated to defense mechanisms, transposon-related proteins, and MDR efflux pumps. Nevertheless, additional interesting features were identified in specific isolates’ genomes. The two isolates of serotype 2 harbored a large gene cluster of Mu-like pro- phage-encoded proteins and other prophage proteins were identified in assemblies ASM16709 (serotype 5) and ASM17859 (serotype 10). The later one also integrated CRISPR-Cas-related proteins, that were also identified in both isolates of serotype 7. In addition, a viral defense system was also annotated in a large gene cluster of strain App6 (GenBank accession number CP029003.1, assembly ASM1735746). DISCUSSION Actinobacillus pleuropneumoniae (APP) causes one of the most economically rele- vant infectious diseases in swine production. Proper serotyping and identification of resistance and virulence profiles are essential to optimize preventive medicine pro- grams based on antimicrobial and vaccine use. Furthermore, genomic comparisons among serotypes provide relevant molecular and epidemiological knowledge to better understand diversity and transmission patterns of this bacterial species across the swine production system. However, to date, this is the first study to include all the available sequencing data from NCBI for APP. Moreover, this comparative analysis includes the first WGS data of APP isolated from Spain. Regarding the epidemiological link between Spanish farms, the phylogenetic tree clustered isolates from breed 1 and breed 2 in separate branches highlighting the pres- ence of different lineages of APP circulating in Spanish farms. As observed, the three isolates from the three different farms supplied by GM1 and M1 during 2018 and 2019 were genetically related (0–1 nucleotide dissimilarity [Table S1]), suggesting that the pathogen is persisting and transmitted between the different production levels in a vertical way. Inversely, isolates from breed 2 that came from the GM2 did not cluster together in the phylogenetic tree. This observation is interesting as it was known that in the multiplication farm M2, as opposite to multiplication farm M3, animals were treated with oxytetracycline for a long period of time due to concomitant problems of leptospirosis (Fraile L, personal communication). The two isolates obtained from two different farms (UG1970008 and UG1970106) and supplied by M2 carried a plasmid and harbored four different resistance genes, including tetracycline resistance, hence, July/August 2022 Volume 10 Issue 4 10.1128/spectrum.01185-22 7 Genomics of Actinobacillus pleuropneumoniae in Pigs Microbiology Spectrum FIG 4 Pangenome analysis representation with anvi'o software. Isolates are colored by serotyped and ordered by presence/absence of gene clusters. (A) Full-pangenome representation including the core, soft-core, shell, and cloud bins. (B) Core representation, including the core and soft-core bins. (C) Accessory representation, including cloud and shell bins. Red and white matrix represents ANI values between isolates. July/August 2022 Volume 10 Issue 4 10.1128/spectrum.01185-22 8 Genomics of Actinobacillus pleuropneumoniae in Pigs Microbiology Spectrum sustaining further the vertical transmission and supporting the evidence of resistance acquisition during or after antimicrobial treatment. On the contrary, isolates from breed 2 from selection farms GM3 and GM4 were phylogenetically related and clus- tered together in the tree (1–128 nucleotide dissimilarities [Table S1]) suggesting peri- odical introductions of the pathogen with no signals of adaptation to the different farm environments. Interestingly, the two isolates from The Netherlands (CVJ13261 and 56153) also appeared closely related to the Spanish isolates from breed 2 (40–114 nucleotide dissimilarities [Table S1]). This association may be explained by the move- ment of breeders across Europe, since The Netherlands is one of the largest exporters of live pigs globally, together with Denmark, Canada, and Germany (26). This same hy- pothesis may explain the association of specific isolates clustering together by country and serotype. For example, some Danish isolates appeared to be related to Australian isolates of serotypes 2 and 6, whereas the Canadian isolate of serotype 7 seems to be related to Chinese and Hungarian isolates of the same serotype. Regardless, the total number of isolates is extremely low to go further with these epidemiological links, but WGS paves the way to carry out wide epidemiological studies when plenty of sequen- ces will be available. APP is relatively understudied using WGS and it hampers the automation of the bio- informatic analysis pipeline. The absence of this pathogen in several databases, such as PointFinder to study point mutations, and the unavailability of the RTX toxin encod- ing genes in the reference virulence factor database (VFDB), made the downstream analyses from assembled genomes more complex and demanding (27, 28). Another challenge for the automation of virulence genes detection was the presence of tandem repeats in the apxIV toxin encoding gene. It is well recognized that these types of repeats lead to sequencing errors that make genome assembly and annotation difficult (29). Therefore, even after setting the parameters of identity at 90% and the coverage at 50%, we were unable to find the apxIV encoding gene in approximately 20% of the isolates, which did not correlate per serotype or per genotype. A more in-depth analy- sis found fractions of this gene located at the end or beginning of contigs and identi- fied the apxIV encoding gene in all APP isolates. These observations associated the inability to detect the gene in some isolates to sequencing errors due to the presence of repetitive regions in the gene. The implementation of in silico PCRs in this analysis allowed the serotyping of unsero- typed isolates uploaded in NCBI, as well as the identification of incorrectly serotyped iso- lates. From these findings, no phylogenetic association was observed between serotypes 7 and 13 in disagreement with previously reported studies (20). In silico PCR demonstrated that strain N273 was incorrectly serotyped and belonged to serotype 7. This result was also supported by the phylogeny tree. Additionally, this technique enabled the proper dis- crimination between serotypes 9 and 11. These serotypes have always been detected with the same PCR primers, despite the fact that the conventional PCR cannot distinguish a sin- gle base deletion and has difficulties with less than 100 bp difference in a .1 kb product (3). The clustering of these two serotypes with serotype 1 in both pangenome reconstruc- tion and phylogenetic tree corresponded with previously reported cross-reactivity between these three virulent serotypes (30, 31). Therefore, we have set the parameters in silico to accurately differentiate serotypes for a proper characterization of isolates. The antimicrobial resistance in silico profiles did not show a clear serotype-specific correlation, as it has been already observed in other studies (32, 33). However, the presence of AMR genes was evidently linked to the presence of plasmids. Many of the resistance genes detected, including aph(30)-Ib, rob-1, floR, and sul2 were located in mobile genetic elements that would facilitate the transmission of the resistance. In fact, the reconstruction of these plasmids showed high homology with those described for pathogens of the Pasteurellaceae family, suggesting that transfer of genetic material between different genera of this family may occur easily. Similarly, the most common identified prophage from the Mannheimia haemolytica species is also a causative agent of a respiratory disease in bovine, and also a pathogen from the Pasteurellaceae family, July/August 2022 Volume 10 Issue 4 10.1128/spectrum.01185-22 9 Genomics of Actinobacillus pleuropneumoniae in Pigs Microbiology Spectrum like the genera Actinobacillus, Pasteurella, and Haemophilus. Interestingly, the two anti- microbial susceptibility patterns identified for the Spanish serotype 11 isolates corre- lated with the antimicrobial resistance in silico profiles: a clear pan-susceptible group and a group resistant to diaminopyrimidine and tetracycline antibiotics. However, for serotype 13 isolates, two in silico identified genes (rob-1 and sul2) did not correspond with the phenotypic susceptible pattern observed in the sensitivity tests, suggesting the nonexpression and noninduction of these genes. Overall, the genomic comparative analyses performed in this study make a signifi- cant contribution to the knowledge of the different serotypes of APP and corroborate toxin profile specificity per serotype. It also highlights the importance of databases being upgraded for automation of the bioinformatic analyses, especially for relatively unknown, albeit relevant, pathogens. Moreover, it increases understanding of phyloge- netic-related serotypes and raises awareness of antimicrobial resistance genes and their potential to be transmitted through mobile genetic elements. Also, the epidemio- logical evidence observed clearly suggests a vertical transmission of these resistance traits within integrated systems, facilitating an epidemiological approach for early treatment of the disease in affected farms. Altogether, the study of the genetic variabil- ity between all available assemblies of APP at the time of the study demonstrated con- siderable differences within serotypes and provided information about prevalence of resistance among this community, strengthening the knowledge for proper surveil- lance and control of this pathogen. MATERIALS AND METHODS Selection of clinical samples, sequencing, assembly, and annotation. Sixteen APP strains were isolated from deceased or diseased pigs showing acute clinical signs of respiratory tract infections. These animals were not exposed to any antimicrobial treatment for at least 15 days before sampling. Selection was based on the availability of the complete epidemiological link between 16 different fattening farms (origin of the isolates), multiplication or mothers’ farm (M), and grandmothers’ farm (GM) detailed in Fig. 1. Two breeders were identified from each selection farm and isolates were collected from 2018 onwards. The DNA from the 16 selected APP isolates was extracted using the DNeasy Ultraclean Microbial Kit (Qiagen) according to the manufacturer’s instructions. Genomes were sequenced using the Illumina MiSeq platform (Illumina Inc., San Diego, CA) and with a MiSeq V3 kit using the 2  250 bp paired end chemistry. Sequencing reads were identified for potential species contamination using Kraken v1.1.1 (34). The trimming step was performed using Trimmomatic v0.39 with a four-base sliding window to cut when the average Phred quality scores dropped below 15. Bases at the start or at the end of the reads with a quality below 3 were also removed and reads shorter than 36 bp were excluded (35). Draft genomes were assembled de novo using the SPAdes assembler v3.14.1 including the module for preliminary read error correction based on Hamming graphs and Bayesian subclustering (36). K val- ues and Phred scores were set to be automatically detected and the mismatch corrector tool BWA was also included. Quality evaluation of assemblies was performed using BUSCO v4.1.4 and QUAST v5.0.2 software (37, 38). Genomes were annotated using Prokka v1.14.6 (39). External data. Additionally, all the assemblies available for APP (taxid:715) at the National Center for Biotechnology Information (NCBI) were downloaded (May 2021) and included for further analyses (40) (Table 1). Quality assessment of these assemblies was also performed with QUAST v5.0.2 and annotated with Prokka v1.14.6 (20, 30). In silico serotyping. Unserotyped genome assemblies downloaded from NCBI were serotyped in sil- ico using the Unipro UGENE in silico PCR tool (41). Already serotyped assemblies were also analyzed. Primer pairs used for specific detection of APP serotypes are described in Bossé, J.T., et al. (3). The num- ber of maximum mismatches was set at 3 bp for both forward and reverse primers and the minimum 39 perfect match was also set at 3 bp. Furthermore, all isolates identified as serotypes 9 or 11 were further analyzed for proper discrimination. This analysis included the 13 Spanish serotype 11 isolates and the APP strains CVJ13261 and 56153, belonging to serotypes 9 and 11, respectively. In silico PCR products were aligned using the UniPro UGENE alignment tool to detect the single base deletion in the final cps gene in serotype 11 reported by Bossé J.T., et al. (3). Variant calling and phylogenetic analyses. Snippy v4.6.0 was used to generate a core genome align- ment based on single nucleotide polymorphisms (SNPs) and insertions/deletions (indels) (42). The ‘snippy- clean_full_aln’ program was used to remove strange characters and the resultant alignment was analyzed for recombination with Gubbins v2.4.1 (43). The core alignment obtained after the removal of the polymor- phic sites was used to generate a phylogenetic tree with IQtree v2.0.3 with 1000 bootstrap replicates and ascertainment bias correction (44). The best-fit substitution model detected was the transversion model with empirical base frequencies. The final phylogenetic tree was visualized with FigTree v1.4.4 (45). Final represen- tation was performed using Phandango v1.3.0 that allowed the association with isolates’ metadata (46). The resulting full and core alignments were also loaded in Unipro UGENE to generate new align- ments using the multiple sequence comparison by log-expectation (MUSCLE) method (41, 47). These July/August 2022 Volume 10 Issue 4 10.1128/spectrum.01185-22 10 Genomics of Actinobacillus pleuropneumoniae in Pigs Microbiology Spectrum alignments were used to construct distance matrixes of both full and core alignments by Hamming dis- similarity and represent them by counts of dissimilarity. Finally, to investigate if there was enough temporal signal in the data to perform phylogenetic mo- lecular clock analysis, both TempEst v1.5.3 and BactDating v1.1 tools were implemented (48, 49). Virulence factors analysis. Presence of virulence factors encoding genes were identified perform- ing BLAST searches with Bandage (50). A database including apxI, apxII, apxIII, and apxIV genes was cre- ated, and serotype-specificity was studied. Gene encoding the ApxIV hemolysin was further analyzed using Unipro UGENE tools for specific sequence patterns (41, 51). Resistance profile, plasmid, and prophage identification. From all annotated assemblies, resist- ance genes were identified using ABRicate v1.0.1 with the CARD database (52, 53). Identification of plas- mids was performed with PLSDB v2021_06_23_v2 using the mash screen strategy with a maximum P-value of 0.1 and a minimum identity of 0.98 (54). Besides, the plasmidSPAdes algorithm v3.14.1 was used to perform plasmid assemblies from the raw reads of the 16 selected APP isolates of this study (55). BLAST search tool in the NCBI database was used to analyze resultant scaffolds for rapid sequence comparison (56). Potential active prophages were located and annotated with Prophage Hunter (accessed August 2021) and genomic islands were predicted with Alien Hunter (accessed June 2021) and visualized with Artemis v18.1.0 (57, 58). Isolation, identification, and antimicrobial susceptibility testing. Clinical specimens isolated from Spanish farms were cultured aseptically, and identification of APP isolates was performed as pre- viously described (6). MIC values were determined using the broth microdilution method (Sensititre, Trek diagnostic Systems Inc., East Grinstead, UK) and following the Clinical and Laboratory Standards Institute (CLSI) (59). The antimicrobials tested included amoxicillin, ceftiofur, doxycycline, enrofloxacin, florfenicol, marbofloxacin, oxytetracycline, sulfamethoxazole/trimethoprim, tiamulin, tilmicosin, tildi- pirosin, and tulathromycin (6). Clinical breakpoints (CB) from CLSI were used to determine antimicrobial susceptibility (59, 60). However, CLSI CB for sulfamethoxazole/trimethoprim and Pasteurellaceae have not been set. Thus, the CLSI CB available for Streptococcus suis (0.5 m g/mL) and sulfamethoxazole/trimetoprim were used in this study. The CB for amoxicillin (0.5 m g/mL) was obtained from the literature and CLSI CB available for tet- racycline (0.5 m g/mL) and enrofloxacin and porcine respiratory pathogens were extrapolated for doxycy- cline, oxytetracycline, and marbofloxacin, respectively (59–61). Pangenome construction. Pangenome analyses were executed following the anvi’o v7 workflow for microbial pangenomics (62, 63). This workflow allowed the identification of gene clusters among the genomes under study. Prior to the pangenome construction, all genome FASTA files were reformatted and converted into an anvi’o contigs database with the ‘anvi-script-reformat-fasta’ and the ‘anvi-gen-con- tigs-database’ programs. The program ‘anvi-run-ncbi-cogs’ was used for gene annotation of the contigs. For constructing the pangenome, the external genomes were first included in a new anvi’o genomes stor- age using the ‘anvi-gen-genomes-storage’ program. Following, the program ‘anvi-pan-genome’ run the pan- genomic analysis on all the stored genomes searching by amino acid sequence similarity with the NCBI’s blastp tool. Additional metadata was added with the ‘anvi-import-misc-data’ program and average nucleotide identity (ANI) was computed with the ‘anvi-compute-genome-similarity’ program using the pyANI tool (64). Finally, the pangenome was visualized in the anvi’o interactive interface with the ‘anvi-display-pan’ program. For further analyses, the full pangenome was split into core and accessory bins, based on the gene clusters frequency across the genomes. Data availability. The genome sequences of the 16 isolates have been deposited in NCBI under the BioProject accession number PRJNA781224. The accession numbers per isolate are listed in Table 1. SUPPLEMENTAL MATERIAL Supplemental material is available online only. SUPPLEMENTAL FILE 1, PDF file, 1.2 MB. ACKNOWLEDGMENTS This study was partially funded by grant RTI2018-095586-B-C22 from the Ministerio de Economía y Competitividad (MINECO), by the CERCA program from Generalitat de Catalunya, and by the AEMPS/PRAN with European funds with reference “Mecanismo de recuperación y resiliencia.” J.G.-M. is a PhD student from the Autonomous University of Barcelona, Biotechnology Program. N.G.-E. and M.M. (FDA) were supported by the FDA Foods Program Intramural Funds. REFERENCES 1. Zimmerman JJ, Karriker LA, Ramirez A, Schwartz KJ, Stevenson GW, Zhang 3. Bossé JT, Li Y, Fernandez Crespo R, Lacouture S, Gottschalk M, Sárközi R, J. 2019. Diseases of Swine, 11th ed. John Wiley & Sons, Inc. Hoboken, NJ. Fodor L, Casas Amoribieta M, Angen Ø , Nedbalcova K, Holden MTG, Maskell 2. VanderWaal K, Deen J. 2018. Global trends in infectious diseases of swine. DJ, Tucker AW, Wren BW, Rycroft AN, Langford PR, BRaDP1T consortium. Proc Natl Acad Sci U S A 115:11495–11500. https://doi.org/10.1073/pnas 2018. Comparative sequence analysis of the capsular polysaccharide loci of .1806068115. Actinobacillus pleuropneumoniae serovars 1–18, and development of two July/August 2022 Volume 10 Issue 4 10.1128/spectrum.01185-22 11 Genomics of Actinobacillus pleuropneumoniae in Pigs Microbiology Spectrum multiplex PCRs for comprehensive capsule typing. Vet Microbiol 220:83–89. 21. Zhan B, Angen Ø, Hedegaard J, Bendixen C, Panitz F. 2010. Draft genome https://doi.org/10.1016/j.vetmic.2018.05.011. sequences of Actinobacillus pleuropneumoniae serotypes 2 and 6. J Bac- 4. Rodrigues da Costa M, Fitzgerald RM, Manzanilla EG, O’Shea H, Moriarty J, teriol 192:5846–5847. https://doi.org/10.1128/JB.00867-10. McElroy MC, Leonard FC. 2020. A cross-sectional survey on respiratory dis- 22. Li G, Xie F, Zhang Y, Wang C. 2012. Draft genome sequence of Actinobacil- ease in a cohort of Irish pig farms. Ir Vet J 73:1–10. https://doi.org/10 lus pleuropneumoniae serotype 7 strain S-8. J Bacteriol 194:6606–6607. .1186/s13620-020-00176-w. https://doi.org/10.1128/JB.01650-12. 5. European Union. Regulation (EU) 2019/6 on veterinary medicinal prod- 23. Donà V, Perreten V. 2018. Comparative genomics of the first and com- ucts and repealing Directive 2001/82/EC. Offical J the European Union 11 plete genome of “Actinobacillus porcitonsillarum” supports the novel spe- December 2018. cies hypothesis. Int J Genomics 2018:5261719. https://doi.org/10.1155/ 6. Vilaró A, Novell E, Enrique-Tarancón V, Balielles J, Allué E, Fraile L. 2020. 2018/5261719. Antimicrobial stewardship for respiratory pathogens in swine. Antibiotics 24. Pereira MF, Rossi CC, de Carvalho FM, de Almeida LGP, Souza RC, de 9:727–711. https://doi.org/10.3390/antibiotics9110727. Vasconcelos ATR, Bazzolli DMS. 2015. Draft genome sequences of six Acti- 7. Loera-Muro A, Angulo C. 2018. New trends in innovative vaccine develop- nobacillus pleuropneumoniae serotype 8 Brazilian clinical isolates: insight ment against Actinobacillus pleuropneumoniae. Vet Microbiol 217:66–75. into new applications. Genome Announc 3:e01585-14. https://doi.org/10 https://doi.org/10.1016/j.vetmic.2018.02.028. .1128/genomeA.01585-14. 8. Stringer OW, Bossé JT, Lacouture S, Gottschalk M, Fodor L, Angen Ø, 25. Bossé JT, Chaudhuri RR, Li Y, Leanse LG, Crespo RF, Coupland P, Holden Velazquez E, Penny P, Lei L, Langford PR, Li Y. 2021. Proposal of Actinoba- MTG, Bazzolli DM, Maskell DJ, Tucker AW, Wren BW, Rycroft AN, Langford PR, cillus pleuropneumoniae serovar 19, and reformulation of previous multi- Peters SE, Weinert LA, Wang J, Luan SL, Maglennon GA, Beddow J, Cuccui J, plex PCRs for capsule-specific typing of all known serovars. Vet Microbiol Terra VS. 2016. Complete genome sequence of MIDG2331, a genetically trac- 255:109021 https://doi.org/10.1016/j.vetmic.2021.109021. table serovar 8 clinical isolate of Actinobacillus pleuropneumoniae. Genome 9. Beck M, Van den Bosch JF, Jongenelen IMCA, Loeffen PLW, Nielsen R, Announc 4:e01667-15. https://doi.org/10.1128/genomeA.01667-15. Nicolet J, Frey J. 1994. RTX toxin genotypes and phenotypes in Actinoba- 26. Food and Agriculture Organisation (FAO). 2020. FAOSTAT: Statistical cillus pleuropneumoniae field strains. J Clin Microbiol 32:2749–2754. database. https://doi.org/10.1128/jcm.32.11.2749-2754.1994. 27. Zankari E, Allesøe R, Joensen KG, Cavaco LM, Lund O, Aarestrup FM. 2017. 10. Kamp EM, Vermeulen TMM, Smits MA, Haagsma J. 1994. Production of PointFinder: a novel web tool for WGS-based detection of antimicrobial Apx toxins by field strains of Actinobacillus pleuropneumoniae and Actino- resistance associated with chromosomal point mutations in bacterial bacillus suis. Infect Immun 62:4063–4065. https://doi.org/10.1128/iai.62.9 pathogens. J Antimicrob Chemother 72:2764–2768. https://doi.org/10 .4063-4065.1994. .1093/jac/dkx217. 11. Klitgaard K, Friis C, Angen Ø, Boye M. 2010. Comparative profiling of the 28. Chen L, Yang J, Yu J, Yao Z, Sun L, Shen Y, Jin Q. 2005. VFDB: A reference transcriptional response to iron restriction in six serotypes of Actinobacil- database for bacterial virulence factors. Nucleic Acids Res 33:D325–D328. lus pleuropneumoniae with different virulence potential. BMC Genomics https://doi.org/10.1093/nar/gki008. 11:698–617. https://doi.org/10.1186/1471-2164-11-698. 29. Tørresen OK, Star B, Mier P, Andrade-Navarro MA, Bateman A, Jarnot P, 12. Schuwerk L, Hoeltig D, Waldmann KH, Valentin-Weigand P, Rohde J. 2021. Gruca A, Grynberg M, Kajava AV , Promponas VJ, Anisimova M, Jakobsen KS, Linke D. 2019. Tandem repeats lead to sequence assembly errors and Sero- and apx-typing of German Actinobacillus pleuropneumoniae field iso- impose multi-level challenges for genome and protein databases. Nucleic lates from 2010 to 2019 reveals a predominance of serovar 2 with regular Acids Res 47:10994–110006. https://doi.org/10.1093/nar/gkz841. apx-profile. Vet Res 52:10–10. https://doi.org/10.1186/s13567-020-00890-x. 30. Rossi CC, Fernandes de Araújo E, Vieira de Queiroz M, Bazzolli DMS. 2013. 13. Lacouture S, Gottschalk M. 2020. Distribution of Actinobacillus pleuro- Characterization of the omlA gene from different serotypes of Actinobacil- pneumoniae (from 2015 to June 2020) and Glaesserella parasuis (from lus pleuropneumoniae: A new insight into an old approach. Genet Mol 2017 to June 2020) serotypes isolated from diseased pigs in Quebec. Ca- Biol 36:243–251. https://doi.org/10.1590/S1415-47572013005000012. nadian Veterinary J 61:1261. 31. Paradis S E, Dubreuil JD, Gottschalk M, Archambault M, Jacques M. 1999. 14. Prado IGDO, Da Silva GC, Crispim JS, Vidigal PMP, Nascimento M, Santana Inhibition of adherence of Actinobacillus pleuropneumoniae to porcine re- MF, Bazzolli DMS. 2020. Comparative genomics of Actinobacillus pleuro- spiratory tract cells by monoclonal antibodies directed against LPS and pneumoniae serotype 8 reveals the importance of prophages in the partial characterization of the LPS receptors. Curr Microbiol 39:313–320. genetic variability of the species. Int J Genomics 2020:9354204. 2020. https://doi.org/10.1007/s002849900465. https://doi.org/10.1155/2020/9354204. 32. Asawa T, Kobayashi H, Mitani K, Ito N, Morozumi T. 1995. Serotypes and 15. Bossé JT, Li Y, Rogers J, Crespo RF, Li Y, Chaudhuri RR, Holden MTG, antimicrobial susceptibility of “Actinobacillus pleuropneumoniae” isolated Maskell DJ, Tucker AW, Wren BW, Rycroft AN, Langford PR, Tucker AW, from piglets with pleuropneumonia. J Vet Med Sci 57:757–759. https:// Peters SE, Weinert LA, Wang J, Luan SL, Maglennon GA, Beddow J, Cuccui doi.org/10.1292/jvms.57.757. J, Terra VS. 2017. Whole genome sequencing for surveillance of antimi- 33. Kim B, Hur J, Lee JY, Choi Y, Lee JH. 2016. Molecular serotyping and anti- crobial resistance in Actinobacillus pleuropneumoniae. Front Microbiol 8: microbial resistance profiles of Actinobacillus pleuropneumoniae isolated 311. https://doi.org/10.3389/fmicb.2017.00311. from pigs in South Korea. Vet Q 36:137–144. https://doi.org/10.1080/ 16. Kilian M, Nicolet J, Biberstein EL. 1978. Biochemical and serological char- 01652176.2016.1155241. acterization of Haemophilus pleuropneumoniae (Matthews and Pattison 34. Wood DE, Salzberg SL. 2014. Kraken: ultrafast metagenomic sequence 1961) Shope 1964 and proposal of a neotype strain. Int J Systematic Bac- classification using exact alignments. Genome Biol 15:R46–12. https://doi teriology 28:20–26. https://doi.org/10.1099/00207713-28-1-20. .org/10.1186/gb-2014-15-3-r46. 17. Pohl S, Bertschinger HU, Frederiksen W, Mannheim W. 1983. Transfer of Hae- 35. Bolger AM, Lohse M, Usadel B. 2014. Trimmomatic: A flexible trimmer for mophilus pleuropneumoniae and the Pasteurella haemolytica-like organism Illumina sequence data. Bioinformatics 30:2114–2120. https://doi.org/10 causing porcine necrotic pleuropneumonia to the genus Actinobacillus (Acti- .1093/bioinformatics/btu170. nobacillus pleuropneumoniae comb. nov.) on the basis of phenotypic and 36. Bankevich A, Nurk S, Antipov D, Gurevich AA, Dvorkin M, Kulikov AS, deoxyribonucleic acid relatedness. Int J Systematic Bacteriology 33:510–514. Lesin VM, Nikolenko SI, Pham S , Prjibelski AD, Pyshkin AV, Sirotkin AV, https://doi.org/10.1099/00207713-33-3-510. Vyahhi N, Tesler G, Alekseyev MA, Pevzner PA. 2012. SPAdes: A new ge- 18. Foote SJ, Bossé JT, Bouevitch AB, Langford PR, Young NM, Nash JHE. 2008. nome assembly algorithm and its applications to single-cell sequencing. J The complete genome sequence of Actinobacillus pleuropneumoniae L20 Comput Biol 19:455–477. https://doi.org/10.1089/cmb.2012.0021. (serotype 5b). J Bacteriol 190:1495–1496. https://doi.org/10.1128/JB.01845-07. 37. Manni M, Berkeley MR, Seppey M, Simão FA, Zdobnov EM. 2021. BUSCO 19. Xu Z, Zhou Y, Li L, Zhou R, Xiao S, Wan Y, Zhang S, Wang K, Li W, Li L, Jin update: novel and streamlined workflows along with broader and deeper H, Kang M, Dalai B, Li T, Liu L, Cheng Y, Zhang L, Xu T, Zheng H, Pu S, phylogenetic coverage for scoring of eukaryotic, prokaryotic, and viral Wang B, Gu W, Zhang XL, Zhu GF, Wang S, Zhao GP, Chen H. 2008. Ge- genomes. Mol Biol Evol 38:4647–4654. https://doi.org/10.1093/molbev/ nome biology of Actinobacillus pleuropneumoniae JL03 an isolate of sero- msab199. type 3 prevalent in China. PLoS One 3:e1450. https://doi.org/10.1371/ 38. Gurevich A, Saveliev V, Vyahhi N, Tesler G. 2013. QUAST: quality assess- journal.pone.0001450. ment tool for genome assemblies. Bioinformatics 29:1072–1075. https:// 20. Xu Z, Chen X, Li L, Li T, Wang S, Chen H, Zhou R. 2010. Comparative doi.org/10.1093/bioinformatics/btt086. genomic characterization of Actinobacillus pleuropneumoniae. J Bacteriol 39. Seemann T. 2014. Prokka: Rapid prokaryotic genome annotation. Bioin- 192:5625–5636. https://doi.org/10.1128/JB.00535-10. formatics 30:2068–2069. https://doi.org/10.1093/bioinformatics/btu153. July/August 2022 Volume 10 Issue 4 10.1128/spectrum.01185-22 12 Genomics of Actinobacillus pleuropneumoniae in Pigs Microbiology Spectrum 40. Schoch CL, Ciufo S, Domrachev M, Hotton CL, Kannan S, Khovanskaya R, Antibiotic resistome surveillance with the comprehensive antibiotic re- Leipe D, McVeigh R, O’Neill K, Robbertse B, Sharma S, Soussov V, Sullivan sistance database. Nucleic Acids Res 48:D517–D525. https://doi.org/10 JP, Sun L, Turner S, Karsch-Mizrachi I. 2020. NCBI Taxonomy: a compre- .1093/nar/gkz935. hensive update on curation, resources and tools. https://doi.org/10.1093/ 54. Galata V, Fehlmann T, Backes C, Keller A. 2019. PLSDB: a resource of com- database/baaa062. plete bacterial plasmids. Nucleic Acids Res 47:D195–D202. https://doi 41. Okonechnikov K, Golosova O, Fursov M, Varlamov A, Vaskin Y, Efremov I, .org/10.1093/nar/gky1050. German Grehov OG, Kandrov D, Rasputin K, Syabro M, Tleukenov T, 55. Antipov D, Hartwick N, Shen M, Raiko M, Lapidus A, Pevzner PA. 2016. Plas- UGENE team. 2012. Unipro UGENE: a unified bioinformatics toolkit. Bioin- midSPAdes: assembling plasmids from whole genome sequencing data. Bio- formatics 28:1166–1167. https://doi.org/10.1093/bioinformatics/bts091. informatics 32:3380–3387. https://doi.org/10.1093/bioinformatics/btw493. 42. Seemann T. 2015. Snippy: Rapid haploid variant calling and core SNP phy- 56. Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ. 1990. Basic local logeny. GitHub. Available at: https://github.com/tseemann/snippy. alignment search tool. J Mol Biol 215:403–410. https://doi.org/10.1016/ 43. Croucher NJ, Page AJ, Connor TR, Delaney AJ, Keane JA, Bentley SD, S0022-2836(05)80360-2. Parkhill J, Harris SR. 2015. Rapid phylogenetic analysis of large samples of 57. Song W, Sun HX, Zhang C, Cheng L, Peng Y, Deng Z, Wang D, Wang Y, Hu recombinant bacterial whole genome sequences using Gubbins. Nucleic M, Liu W, Yang H, Shen Y, Li J, You L, Xiao M. 2019. Prophage Hunter: an Acids Res 43:e15-e15. https://doi.org/10.1093/nar/gku1196. integrative hunting tool for active prophages. Nucleic Acids Res 47: 44. Minh BQ, Schmidt HA, Chernomor O, Schrempf D, Woodhams MD, Von W74–W80. https://doi.org/10.1093/nar/gkz380. Haeseler A, Lanfear R, Teeling E. 2020. IQ-TREE 2: new models and effi- 58. Vernikos GS, Parkhill J. 2006. Interpolated variable order motifs for identi- cient methods for phylogenetic inference in the genomic era. Mol Biol fication of horizontally acquired DNA: Revisiting the Salmonella pathoge- Evol 37:1530–1534. https://doi.org/10.1093/molbev/msaa015. nicity islands. Bioinformatics 22:2196–2203. https://doi.org/10.1093/ 45. Rambaut A. 2018. FigTree v. 1.4.4. Github. Available at: https://github bioinformatics/btl369. .com/rambaut/figtree. 59. Lubbers BV, Papich MG, Schwarz S, Bowden R, Diaz-Campos DV, Fielder 46. Hadfield J, Croucher NJ, Goater RJ, Abudahab K, Aanensen DM, Harris SR. M, Langston C, Li XZ, Martinez MN, Miller C, Pallotta C, Shryock TR, Simjee 2018. Phandango: an interactive viewer for bacterial population genomics. S, Sinnott-Stutzman V, Sweeney MT, Traczewski MM, Trott D, Yan SS. Bioinformatics 34:292–293. https://doi.org/10.1093/bioinformatics/btx610. 2018. Performance Standards for Antimicrobial Disk and Dilution Suscep- 47. Edgar RC. 2004. MUSCLE: Multiple sequence alignment with high accu- tibility Tests for Bacteria Isolated from Animals. 4th ed CLSI supplement racy and high throughput. Nucleic Acids Res 32:1792–1797. https://doi VET08. Clinical and Laboratory Standards Institute. .org/10.1093/nar/gkh340. 60. Weinstein MP, Lewis Ll JS, Bobenchik AM, Campeau S, Cullen SK, Galas 48. Rambaut A, Lam TT, Carvalho LM, Pybus OG. 2016. Exploring the temporal MF, Gold H, Humphries RM, Kirn JT, Limbago B, Mathers AJ, Mazzzuli T, structure of heterochronous sequences using TempEst (formerly Path-O- Gen). Virus Evol 2:vew007. https://doi.org/10.1093/ve/vew007. Satlin M, Schuetz AN, Simner PJ, Tamma PD. 2020. Performance Standards 49. Didelot X, Croucher NJ, Bentley SD, Harris SR, Wilson DJ. 2018. Bayesian for Antimicrobial Susceptibility Testing. 30th ed CLSI supplement M100. inference of ancestral dates on bacterial phylogenetic trees. Nucleic Acids Clinical and Laboratory Standards Institute. Wayne, PA. Res 46:e134-e134. https://doi.org/10.1093/nar/gky783. 61. Schwarz S, Böttner A, Goosens L, Hafez HM, Hartmann K, Kaske M, 50. Wick RR, Schultz MB, Zobel J, Holt KE. 2015. Bandage: Interactive visualization Kehrenberg C, Kietzmann M, Klarmann D, Klein G, Krabisch P, Luhofer G, of de novo genome assemblies. Bioinformatics 31:3350–3352. https://doi Richter A, Schulz B, Sigge C, Waldmann KH, Wallmann J, Werckenthin C. .org/10.1093/bioinformatics/btv383. 2008. A proposal of clinical breakpoints for amoxicillin applicable to por- 52. Schaller A, Kuhn R, Kuhnert P, Nicolet J, Anderson TJ, Maclnnes JI, Segers cine respiratory tract pathogens. Vet Microbiol 126:178–188. https://doi RPAM, Frey J. 1999. Characterization of apxIVA, a new RTX determinant of .org/10.1016/j.vetmic.2007.06.023. Actinobacillus pleuropneumoniae. Microbiology 145:2105–2116. https:// 62. Eren AM, Esen OC, Quince C, Vineis JH, Morrison HG, Sogin ML, Delmont doi.org/10.1099/13500872-145-8-2105. TO. 2015. Anvi’o: An advanced analysis and visualization platform for 52. Torsten Seemann BG. 2020. ABRicate. Github. Available at: https://github ’omics data. PeerJ 3:e1319. https://doi.org/10.7717/peerj.1319. .com/tseemann/abricate. 63. Delmont TO, Eren EM. 2018. Linking pangenomes and metagenomes: 53. Alcock BP, Raphenya AR, Lau TTY, Tsang KK, Bouchard M, Edalatmand A, The Prochlorococcus metapangenome. PeerJ 6:e4320–23. https://doi.org/ Huynh W, Nguyen AV, Cheng AA, Liu S, Min SY, Miroshnichenko A, Tran 10.7717/peerj.4320. HK, Werfalli RE, Nasir JA, Oloni M, Speicher DJ, Florescu A, Singh B, Faltyn 64. Pritchard L, Glover RH, Humphris S, Elphinstone JG, Toth IK. 2016. M, Hernandez-Koutoucheva A, Sharma AN, Bordeleau E, Pawlowski AC, Genomics and taxonomy in diagnostics for food security: soft-rotting Zubyk HL, Dooley D, Griffiths E, Maguire F, Winsor GL, Beiko RG, Brinkman enterobacterial plant pathogens. Anal Methods 8:12–24. https://doi.org/ FSL, Hsiao WWL, Domselaar GV, McArthur AG. 2020. CARD 2020: 10.1039/C5AY02550H. July/August 2022 Volume 10 Issue 4 10.1128/spectrum.01185-22 13