Genomic characterization of Actinobacillus pleuropneumoniae serovars 5, 8, and 15 strains from diseased pigs in eastern Chinese provinces

ABSTRACT

Actinobacillus pleuropneumoniae causes porcine infectious pleuropneumonia in pigs. We aimed to characterize the phenotypic and genomic features of three A. pleuropneumoniae strains from clinical cases in eastern Chinese provinces. The serovar 5 strain ZJNH2023 was more pathogenic than strains AH2020 and ZJXS2022 in a murine model and was resistant to multiple antimicrobials. The core genome SNP (single nucleotide polymorphism) tree indicates that the three isolates are clustered with serovars 5, 8, and 15 strains of archived genomes. They harbor plasmids conferring resistance to florfenicol and are of substantial genome diversity, having more prophages, genomic islands (GIs), and antimicrobial resistance genes (ARGs) than the strains of corresponding serovars from other studies. The capsule-related gene clusters in strains AH2022 and ZJXS2022 are different from ZJNH2023 and contain an ISApl1 family transposase between the cps and cpx loci. The serovar 5 strain ZJNH2023 has a full set of ApxI genes, Apa1/Apa2, intact flp family genes related to Flp pilus assembly, and a full set tadABCD genes related to adherence, while strains ZJXS2022 and AH2022 carry ApxIII gene set, lack ApxIAC genes and Apa1/Apa2, and do not have intact flp family genes. Thus, we conclude that possession of the cytotoxic ApxI gene set and those involved in adhesion contributes to higher pathogenicity of the serovar 5 strain ZJNH2023. Distinct GIs and floR-containing plasmids in these strains might have been involved in multiple resistance and horizontal transfer of ARGs on the pig farms.

Introduction

Actinobacillus pleuropneumoniae (App) is a significant contributor to the porcine respiratory disease complex (PRDC), causing significant economic losses to the global pig industry [1,2]. This bacterium has 19 serovars identified so far with notable differences in virulence [3–5]. There are also differences in the geographic distribution of App serovars [5]. Serovars 5, 7, and 8 strains are more prevalent in North America [6,7]. Serovar 8 strains are predominant in the UK [8]. Strains in other European countries, such as Germany and Hungary, and Japan are dominated by serovar 2 [9–11]. Serovars 1, 3, 4, 5, and 7 strains were reportedly predominant in China [12]. However, a recent study indicates that serovar 8 is more prevalent in App isolates in Zhejiang, China, followed by serovars 5, 7, and 15 [13].

App is able to produce four related exotoxins, ApxI, II, III, and IV which belong to the RTX (Repeats in Toxin) family and are associated with virulence. The toxin gene (apx) patterns differ among different strains. A given serovar usually has a distinct apx pattern. The severity of infection with different serovars varies considerably, with some serovars being associated with high mortality and others with no symptoms [1,2,14].

Genomic sequencing of bacterial pathogens is useful for the identification of presumed virulence factors among a diverse range of strains by comparative genomics, elucidation of strain or serovar diversity, horizontal gene transfer, etc. The first reports in 2008 of genome sequencing of App strains JL03 (serovar 3) and L20 (serovar 5b), by Xu et al. (2008) and Foote (2008) [12,15], respectively, were followed by the same Chinese group in 2010 in sequencing App strains of serovars 1, 2, 4, 6, 9, 10, 11, 12, and 13 from Australasian Pig Institute, Australia, using the then high-throughput 454 GS FLX platform [16]. With the development of next-generation sequencing technologies, App strains of more serovars or from different countries have been recently sequenced, including the newly described serovar 19 strains [11,17–19]. However, recent Chinese App strains have not been compared from genomic perspectives.

Moreover, serovars and genomic structures of App isolates in Zhejiang and Anhui, two southeastern Chinese coastal provinces, are largely unknown. During the period from January 2022 to December 2023, we received 137 lung samples from clinical cases of PRDC, among them, cases of mixed infection with App account for 9%. The other pathogenic bacteria found in mixed infections included Glaesserella parasuis, Streptococcus suis, Mycoplasma hyopneumoniae, or Pasteurella multocida. We recovered three App strains which were PCR-typed as serovars 5, 8, and 15. These strains were subjected to genomic sequencing as well as testing on antimicrobial susceptibility and virulence in mice. Our aims were to determine if there are differences in pathogenicity and antimicrobial resistance of these recent App isolates in Zhejiang and Anhui, China, and to explore if there are distinct features in the genomic structures related to these phenotypes or as compared to the strains of corresponding serovars from other studies.

Materials and methods

Ethics statement

The animal experiment for virulence estimation in the murine model was reviewed and approved by the Animal Ethics Committee of the Zhejiang Sci-Tech University (Permission No.20230511-04). The housing and care of experimental animals were carried out in strict accordance with the China National Standards “Laboratory animals – General Code of Animal Welfare” (GB/T42011-2022). All experimental procedures were carried out in compliance with the ARRIVE guidelines.

Bacterial strains, media and growth conditions

The 137 lung samples aseptically collected by field veterinarians and submitted to our laboratory in 2022 and 2023 were from clinical cases of PRDC pigs which were humanely euthanized for necropsy due to disease severity by intramuscular premedication with ketamine hydrochloride (20 mg/kg) for sedation followed by exsanguination via transection of the axillary vessels in accordance with the Animal Ethics Guidelines of the Chinese Veterinary Medical Association. Of these samples, 15 were found infected with A. pleuropneumoniae by preliminary identification and 12 of them had mixed infections with other bacterial or viral pathogens (Tab. S1). A total of three App strains ZJNH2023 and ZJXS2022 (isolated from Ninghai and Xiaoshan, respectively, in Zhejiang Province, China) and AH2022 (from Dangshan, Anhui Province, China) were chosen as they represented single infection with apparent respiratory symptoms and hemorrhagic lung lesions. The strains were biochemically identified by a recommended protocol [20] and further typed by multiplex PCR as serovars 5 (ZJNH2023), 8 (AH2022), and 15 (ZJXS2022) [21]. The strains were kept at −70°C as frozen stocks in tryptic soy broth (TSB, Qingdao Hope Bio-Tech, China) containing 25% glycerol. Before use, the stock cultures were streaked on the tryptic soy agar (TSA, Qingdao Hope Bio-Tech) plate supplemented with 100 µg/mL of nicotinamide adenine dinucleotide (NAD, Genview, USA) and 10% bovine serum, and incubated at 37°C for 24 h. A single colony of each strain from the plates was inoculated into TSB for subculture at 37°C for 6–8 h.

Extraction of genomic DNA and genome sequencing

Genomic DNA was extracted from the above strains using the STE method (sodium chloride-Tris-HCl-EDTA method), and the concentration was quantified with the Qubit®2.0 fluorometer (ThermoFisher Scientific, USA). The DNA samples were placed in a styrofoam box with dry ice and sent for sequencing with the Nanopore PromethION 48 (Oxford Nanopore Technologies, UK) and NovaSeq PE150 (Illumina, USA) platforms [22] at the Beijing Novogene Bioinformatics Technology Co. Ltd. (Beijing, China).

Identification of presumed plasmids

Initial findings of the floR gene in the three strains with resistance phenotype to florfenicol and presence of a plasmid in the strain ZJXS2022 from the sequencing data led us to examine the types of presumed plasmids in our strains: (1) PlasmidSPAdes v4.2 was used to obtain potential floR-containing plasmid contigs in the original reads for the strains ZJXS2022, ZJNH2023, and AH2022. These contigs were further processed with MOB-Suite v3.1.9 to obtain good quality sequences (reconstruction using MOB-Recon function) and plasmid types (using MOB-Typer function). These presumed plasmid sequences were subjected to further BLAST in the NCBI database. (2) PCR using primer pairs targeting different fragments of the plasmid pIV86 with overlaps of one fragment to the other as well as the plasmid circle (Fig. S1) were used for amplification of the target fragments from the plasmid DNA extracts of the strains ZJXS2022 and AH2022 (FastPure Plasmid Mini Kit, Vazyme, Nanjing, China). For the putative plasmid in the strain ZJNH2023 in doubt, the plasmid extract was subjected to re-sequencing on the PromethION 2 Solo with R10.4.1 Chip at Zhejiang Youkang Biotech Co. Ltd., Hangzhou, China, and tested for circular plasmid by PCR using specific primer pairs (Fig. S1).

Genome assembly

The NanoPlot software (https://github.com/wdecoster/NanoPlot, v1.29.1) was used with a threshold Q > 7 to perform quality control on the sequencing data [23]. The tool Unicycler (https://github.com/rrwick/Unicycler) (v0.4.8) was used to conduct hybrid assembly from the PE150 read sets where it functions as a SPAdes-optimizer and from Nanopore long-read sets where it runs on a miniasm + Racon pipeline [24]. Chromosomal sequences or plasmid sequences were distinguished from the assembled sequences based on alignment and sequence length.

Bioinformatics analysis

Genomic characteristics of the three new strains were compared with other strains from NCBI database (two from China with known serovars, and twenty-five from other countries covering all 19 serovars, searched in July, 2024). The App strain ATCC27088 (S4074) was used as a reference genome (Accession No. CP030753.1). Core genome single nucleotide polymorphisms (core SNPs) were identified by Snippy v4.4.4 as described in previous publications [25–27] and aligned to build a maximum-likelihood phylogenetic tree (1000 bootstraps) using IQ-TREE v1.6.12 [28] with the best model TVM+F+ASC+R3.

To predict the gene functions, two databases were used: COG (Clusters of Orthologous Groups), and Swiss-Prot [29]. A BLAST search of the whole genome was carried out against these two databases with an E-value of less than 1e-5 and a minimum percentage of alignment length of more than 40%. Pathogenicity and antimicrobial resistance genetic determinant analyses included VFDB (Virulence Factors of Pathogenic Bacteria) [30], CARD (Comprehensive Antibiotic Research Database), and CGE Resfinder databases [31,32] (similarity >90%, coverage >60%) as previously described [33]. Individual search on the genomes of the three strains was conducted for the genes related to capsule synthesis and export in App strains [17,22].

Genome component prediction involved ORFs, genomic islands (GI), and prophages. ORFs were annotated by Prokka, and then summarized by Roary [34,35]. The GIs were predicted using the IslandPath-DIOMB program [36]. Prophage prediction was carried out using PHAST (http://phast.wishartlab.com/). The BRIG software was used to display the locations of predicted GIs with coding genes in the genome [37].

The plasmid replicons were predicted using MOB-Typer and Plasmidfinder (identity > 80%, coverage > 80%) [31,38]. The GIs, prophages, antimicrobial resistance genes (ARGs), and virulence factors with tree were visualized by R-studio v1.1 with R v4.2.1 and R packages according to the recent publication [33]. PlasmidSPAdes v4.2 (https://usegalaxy.eu/) was used to obtain potential plasmid contigs which were further processed with MOB-Suite v3.1.9 to obtain good quality sequences and plasmid types [31,38]. These plasmid sequences containing floR were subjected to further BLAST in the NCBI database.

Antimicrobial susceptibility testing

Broth micro-dilution method was used to examine the resistance of App strains to antimicrobials in the Veterinary Fastidious Medium (VFM) containing Mueller Hinton broth, yeast extract, and lysed horse blood according to the Clinical and Laboratory Standards Institute (CLSI) guideline [39]. Antimicrobials ampicillin, ceftiofur, tiamulin, enrofloxacin, florfenicol, sulfamethoxazole, tetracycline, tilmicosin, chloramphenicol, and kanamycin (Shanghai Macklin Biochemical Co., Ltd., China) were used and serially diluted to eight working concentrations. The final bacterial concentration was around 5 × 10⁵ CFU/mL in the wells. The mixtures were incubated at 35°C for 20–24 h, and then the microplates were read in a spectrophotometer (BioTek Epoch, USA) at 600 nm. Minimum inhibition concentration (MIC) was defined as the lowest concentration that completely inhibited the growth of the tested bacterial isolates. The breakout points of resistance were based on CLSI [39].

Virulence estimation in a murine model

To investigate the pathogenicity of the isolated strains, 25 female BALB/c mice of specific-pathogen-free (SPF, Shanghai SLAC Laboratory Animal Co. Ltd., China) grade aged between 9 and 10 weeks were divided randomly into five groups for each strain: four experimental groups were given four graded amounts of viable bacterial cells: 8 × 10⁵ to 1 × 10⁸ CFU (strain ZJNH2023), 4 × 10⁶ to 5 × 10⁸ CFU (strain AH2022) and 8 × 10⁶ to 1 × 10⁹ CFU (strain ZJXS2022), and one group served as sham control. The bacteria were revived on a supplemented TSA plate at 37°C for 18 h, and Several colonies from the pure cultures of each strain on the agar plates were transferred to sterilized saline to adjust the OD₆₀₀ to 0.4–0.5. The bacteria were then cultured in the supplemented TSB medium at a ratio of 1:100. The bacterial cultures at the middle logarithmic phase were used to prepare inocula of different bacterial populations for intraperitoneal injection. Sterilized saline was utilized for the sham control mice. Each group of mice was continuously observed for 7 d after injection for clinical signs and mortality. LD₅₀ was determined using the Karber method [40]. The mice survived the experiment were humanely euthanized via gradual CO₂ displacement following the guidelines outlined by the Institutional Animal Ethics Committee of the Zhejiang Sci-Tech University.

Results

Virulence to mice and antimicrobial resistance profiles

Of the three clinical isolates, the serotype 5 strain ZJNH2023 was most pathogenic to mice, while the other two strains were less pathogenic with LD₅₀ of about 6–7 folds less than ZJNH2023 (Table 1). All three isolates were susceptible to ceftiofur, tiamulin, and enrofloxacin, but resistant to florfenicol, chloramphenicol, tetracycline, and kanamycin. The strain ZJXS2022 was resistant to ampicillin, while the other two were susceptible. To tilmicosin, ZJXS2022 was susceptible, but the other two were resistant (Table 2).

Table 1.

Virulence evaluation of three Actinobacillus pleuropneumoniae clinical strains in mice.

Strains	Serotype	Inoculum ranges (cfu/mouse)	LD50
ZJNH2023	5	4.43 × 105 − 5.20 × 107	1.13 × 107
AH2022	8	1.27 × 106 − 1.33 × 108	7.37 × 107
ZJXS2022	15	3.27 × 106 − 3.70 × 108	8.29 × 107

Table 2.

Susceptibility of Actinobacillus pleuropneumoniae strains to different antimicrobials.

		Actinobacillus pleuropneumoniae strains*
		ZJNH2023			AH2022			ZJXS2022
Antimicrobials	Resistance cutoff	MIC	Interp		MIC	Interp		MIC	Interp
Ampicillin	2	$\le$0.25	S		32	R		$\le$0.25	S
Ceftiofur	8	$\le$0.25	S		$\le$0.25	S		$\le$0.25	S
Tiamulin	32	4	S		$\le$2	S		$\le$2	S
Enrofloxacin	1	$\le$0.06	S		0.25	S		$\le$0.6	S
Florfenicol	8	32	R		32	R		32	R
Chloramphenicol	8a	32	R		32	R		32	R
Sulfamethoxazole	N/Ab	16	/		1	/		4	/
Tetracycline	2	16	R		32	R		32	R
Tilmicosin	32	>256	R		32	R		$\le$2	S
Kanamycin	8c	>64	R		>64	R		>64	R

*MIC, minimum inhibitory concentration (μg/mL); Res, resistance level (μg/mL); and Interp, interpretation of the results: S = sensitive and R = resistant.

^aBased on other bacteria in dogs or cats (it is not allowed in food animals) [39].

^bNot available for interpretation.

^cBased on the proposed cutoff value for amikacin and/or gentamicin on other bacteria [39].

Sequencing data quality and genome assembly

The genomes of three App strains were fully sequenced. NanoPlot analysis indicates that the total bases are 1140–1,670 Mb from the numbers of reads between 120,670 and 201,528 with an average length of reads from 8181 to 11,375 bp (Tab. S2). The reads were of good quality shown with the mean read quality from 13.7 to 16.2 and the N50 read length between 11,287 and 14,122 bp (Fig. S2) [41]. The NovaSeq reads were also of good quality with the clean data Q20 and Q30 above 91.56% (Tab. S3). The chromosome sequences of these three strains and those of corresponding serovars from other studies were then assembled into circular genomes showing the GC content, percent identity, and genomic islands (with coding genes, etc.) (Figs. S3-S5). The genome size is 2.35 Mb, 2.30 Mb, and 2.38 Mb for strains ZJNH2023, AH2022, and ZJXS2022, respectively, with the GC content around 41% (Tab. S2). The numbers of coding sequences (CDS, protein-coding) are 2163 (ZJNH2023), 2141 (AH2022), and 2269 (ZJXS2022), 65 to 193 CDSs above the average of 2076 (1,961–2,209) of the twenty-seven strains from other studies [42–44] (Table 3).

Table 3.

Total lengths and numbers of coding sequences in the genomes of thirty Actinobacillus pleuropneumoniae strains.

Strains	Country§	Sero*	Total length	CDS#	Accession No.	References§
ZJNH2023	China	5	2,350,537	2163	CP143255	This study
AH2022	China	8	2,301,308	2141	CP141949	This study
ZJXS2022	China	15	2,415,409	2269	CP141951	This study
KL16	South Korea	1	2,365,505	2209	CP022715	N/A
ATCC27088 (S4074)	Argentina	1	2,318,657	2081	CP030753	Donà et al., 2018
NCTC10976	Denmark	2	2,325,526	2147	LR134515	Foote et al., 2008
P1875	Switzerland	2	2,309,071	2047	CP079921	Donà et al., 2022
S1536	Switzerland	2	2,282,693	2041	CP031875	Donà et al., 2022
ORG1224	NA	3	2,245,829	1997	CP031854	Donà et al., 2022
JL03	China	3	2,242,062	2036	CP000687	Xu et al., 2008
S1421	Switzerland	3	2,235,635	1980	CP031874	Donà et al., 2022
M62	USA	4	2,335,268	2118	CP031873	Donà et al., 2022
K17	USA	5a	2,284,762	2054	CP069797	Donà et al., 2022
L20	USA	5b	2,274,482	2012	CP000569	Foote et al., 2008
APP6	China	5	2,409,072	2160	CP026009	N/A
Femo	Denmark	6	2,409,565	2196	CP069796	Donà et al., 2022
WF83	Canada	7	2,312,414	2084	CP031869	Donà et al., 2022
AP76	Canada	7	2,345,435	2142	CP001091	Xu et al., 2010
MIDG2331	UK	8	2,337,633	2174	LN908249	Bossé et al., 2016
405	Ireland	8	2,321,392	2098	CP031866	Donà et al., 2022
CVJ13261	Netherlands	9	2,324,821	2085	CP031865	Donà et al., 2022
D13039	NA	10	2,310,485	2058	CP031864	Donà et al., 2022
56,153	Netherlands	11	2,324,505	2089	CP031863	Donà et al., 2022
8329	NA	12	2,252,295	2018	CP031862	Donà et al., 2022
N273	NA	13	2,308,478	2066	CP031861	Donà et al., 2022
3906	NA	14	2,223,843	1970	CP031860	Donà et al., 2022
HS143	NA	15	2,240,110	1996	CP031859	Donà et al., 2022
A-85/14	Hungary	16	2,364,304	2145	CP069795	Donà et al., 2022
16,287–1	NA	17	2,300,190	2080	CP031856	Donà et al., 2022
7,311,555	NA	18	2,214,657	1961	CP031855	Donà et al., 2022

^§N/A: not available.

*Serovars.

^#Number of protein-coding sequences.

Genetic relationship with genomes of other A. pleuropneumoniae strains

The core genomes of the three strains in this study and those downloaded from GenBank databases (Table 3) were analyzed for SNPs. There are 61,952 core gSNPs in the genomes of the tested strains. Core gSNP tree reveals that three strains in this study are clustered with serovars 5, 8, and 15 strains (Figure 1), which are identical to those by multiplex PCR typing [21].

Figure 1.

Phylogenetic relationship of the Actinobacillus pleuropneumoniae strains and their major genomic characteristics.

An unrooted maximum likelihood phylogenetic tree constructed using SNPs inside core genome sequences of 30 A. pleuropneumoniae strains with complete genome sequences. The strains ZJNH2023, AH2022, and ZJXS2022 in this study are marked in the tree by symbol “*” in the middle part. The scale bar represents the single nucleotide substitutions. ATCC27088 was chosen as the reference strain. Name, serovar, and origin country of each strain are marked in the format “Name_serovar_country abbreviation” near the corresponding tree node. Columns of the heatmap present distributions of serovar (column 1), number of virulence factor (VF, column 2), number of antimicrobial resistance genes (ARGs, column 3), number of prophages (column 4), and number of genomic islands (column 5). The number of VF genes or other targets for individual strains are categorized by different colors.

Prophages and genomic islands

There are 35 types of prophages, including those incomplete, found in 30 App strains examined. The prophages vary in numbers from 1 to 7 per strain, and differ within the three strains of this study and between our strains and those from other studies. Of the three strains, ZJNH2023 has more prophages (7 incomplete) than AH2022 (5 incomplete) and ZJXS2022 (4 with 2 incomplete and 2 complete – Entero_Mu and Mannhe_vB_MhS_587AP2) (Tab. S4). There are more prophages in the strains of this study than those of the corresponding serovars from other studies: only 1 to 2 prophages in the other two serovar 5 strains; 2 to 3, in the other two serovar 8 strains; and one incomplete, in the serovar 15 strain HS143 (Tab. S4).

Of all the GIs listed (Tab. S5), the Serovar5_GI-2, Serovar8_GI-2, and Serovar15_GI-4 are shared across the strains of the three serovars with similar genetic structures, while there are four shared GIs in the strains between two serovars (Serovars 5 and 15, and Serovars 8 and 15) as shown by identical uppercase letters (Tab. S5). However, the strain ZJXS2022 has higher GI carriage (n = 8) than the other two strains of this study or those of other studies (n = 4–5) (Figs. S3-S5) and possesses three distinct GIs (Serovar15_GI-2, _GI-3, and _GI-5) and two intact prophages (Entero_Mu and Mannhe_vB_MhS_587AP2 that contain GI-2 and GI-5, respectively) that are absent in the other serovar 15 strain HS143 (Tab. S5, Fig. S5). For the strain AH2022, there is a distinct GI (serovar8_GI-1) that is not present in other strains of the same serovar (MIDG2331 and 405). The strains ZJNH2022 and APP6 (another Chinese serovar 5 strain) possess two GIs (Serovar5_GI-1 and _GI-4) that are not present in other serovar 5 strains (K17 and L20) (Tab. S5). Serovar5_GI-1 in ZJNH2023 carries two flagellar biosynthesis transcription regulator genes flhCD and a trimethoprim ARG dfrA6, and the serovar5_GI-4 unique for the two Chinese serovars 5 strains contains four tetracycline ARGs, including a tetB and three incomplete genes tetA, tetD and tetR (Fig. S3).

Profiles of major virulence factors

Of the four RTX toxins, ApxI and ApxIII are highly cytotoxic while ApxII is moderately toxic, and ApxIV probably involved in tissue damage is mainly expressed in vivo [2]. Table 4 shows that the three strains of this study and those of other studies carry the apxIICA and apxIVA genes. The serovar 5 strain ZJNH2023 and those of the same serovar from other studies, though lacking apxIIICABD, contain more virulence genes than strains ZJXS2022 and AH2022: a full set of ApxI genes, Apa1 and Apa2 encoding two autotransporter adhesins involved in colonization and virulence [45] as well as the tad locus genes (flp1/2, flpBCD, rcpA, and tadABCD) involved in Flp pilus assembly and host cell adhesion [46] (Figure 1, Table 4 and S6), which is consistent with higher virulence of the strain ZJNH2023 shown as having the lowest LD₅₀ value (Table 1). However, strains ZJXS2022 and AH2022 in this study do not carry apa1/apa2 and tad locus genes that are otherwise present in the serotypes 8 and 15 strains of other studies (Table 4).

Table 4.

Major virulence genes in serovars 5, 8, and 15 strains of this study and those of other studies.

Virulence genes	Serotype 5		Serotype 8		Serotype 15
	ZJNH2023	Others*	AH2022	Others*	ZJXS2022	Others*
apxI/CABD	++++	++++	- -++	- -++	- -++	- -++
apxII/CA	++	++	++	++	++	++
apxIII/CABD	- - - -	— - - - -	++++	++++	++++	++++
apxIVA	+	+	+	+	+	+
apa1	+	+	–	+	–	+
apa2	+	+	–	+	–	+
flp1	+	+	–	+	–	–
flp2	+	+	–	+	–	+
flp/BCD	+++	+++	- - -	+++	- - -	+++
tad/ABCD	++++	++++	- - - +	++++	- - - +	++++

*The strains of these corresponding serovars from other studies listed in Table 3; “+” and “-” mean presence and absence, respectively.

Genetic organization of the gene clusters related to capsule polysaccharide synthesis and transportation between ydeN and modF is quite different among the three strains, but generally consistent with those of corresponding serovars 5, 8, and 15 [21]. However, there is the insertion of an IS30-like element ISApl1 family transposase between cps and cpx clusters in the strain AH2022 (within the serovar8_GI-4) and upstream of cps15A in the strain ZJXS2022 (within the serovar15_GI-7) (Figure 2; Fig. S4 and S5), which are absent in other strains with sequenced genomes of corresponding serovars, including the new serovar 5 strain ZJNH2023.

Figure 2.

Genetic organization of capsule biosynthesis loci of Actinobacillus pleuropneumoniae strains ZJNH2023, AH2022, and ZJXS2022.

Antimicrobial resistance genes

There are 10 types of ARGs detected in the three strains of this study, 4 for strain ZJNH2023, 9 for AH2022, and 7 for ZJXS2022, which are far more than those of other studies (1 to 3 ARGs) (Table 5). The GIs of strains AH2022 (serovar8_GI-1) and ZJXS2022 (serovar15_GI-3) are the major carriers of multiple genes conferring resistance to aminoglycosides (aph(3’”)-Ib, aph(6)-Id, and aph(3”)-Ia), tetracycline and sulfonamide (sul2) (Tab. S5, Figs. S4 & S5). Tetracycline ARGs, especially the complete tet(B) and incomplete tet(R) genes, are present in the GIs of all the three new strains (Fig. S3-S5), but the incomplete tet(C) and tet(D) genes are present only in the serovar5_GI-4 of ZJNH2023 (Fig. S3). The trimethoprim-resistant gene dfrA6 or the ampicillin-resistant gene bla_ROB-1 is present only in the GI-1 of the strain ZJNH2023 or AH2020, respectively (Figs. S3 & S4).

Table 5.

Annotations of antimicrobial resistance genes in Actinobacillus pleuropneumoniae strains from this study and those of the same serovars of other studies.

Strains	Gene_ID1	Subject_ID2	ARG ID3	Other strains4	Antimicrobial_Resistance
ZJNH2023	V1209_00210	Z86002	dfrA6	-/-	Trimethoprim
(Serovar 5)	V1209_09895	AF326777	tet(B)	-/+	Tetracycline
	V1209_02470	O85735	rsmAe	+/+	Chloramphenicol, florfenicol
	Plasmid	AF118107	floRb	-/+	Chloramphenicol, florfenicol
AH2022	VN992_04300	AF024602	aph(3’’)-Ib	–	Streptomycin
(serovar 8)	VN992_04285	EF015636	aph(3’)-Ia	–	Kanamycin
	VN992_04295	M28829	aph(6)-Id	–	Streptomycin
	VN992_04275	AF022114	blaROB-1	–	Ampicillin
	VN992_04305	FJ197818	sul2	+	Sulfisoxazole
	VN992_04260	AF326777	tet(B)	+	Tetracycline
	VN992_04320	Q54528	estTa	–	Tilmicosin
	VN992_02090	O85735	rsmAa	+	Chloramphenicol, florfenicol
	Plasmid	AF118107	floRc	–	Chloramphenicol, florfenicol
ZJXS2022	VOA82_04545	AP000342	tet(B)	–	Tetracycline
(serovar 15)	VOA82_04555	AY034138	sul2	–	Sulfonamide
	VOA82_04560	AF321551	aph(3’’)-Ib	–	Streptomycin
	VOA82_04565	AB109805	aph(6)-Id	–	Streptomycin
	VOA82_04575	EF015636	aph(3’)-Ia	–	Kanamycin
	VOA82_03475	O85735	rsmAa	+	Chloramphenicol, florfenicol
	Plasmid	AF118107	floRc	–	Chloramphenicol, florfenicol

¹Gene_ID: the “locus_tag” from the Genbank format file of corresponding strains in the NCBI Nucleotide database.

²Subject_ID: the accession numbers of the best matched reference sequences in NCBI.

³ARG ID refers to the names of antimicrobial resistance genes used in Resfinder or CARD databases.

⁴The strains of the same serovars from other studies listed in Table 3. Serovar 5: “x/y,” other country strains/another Chinese strain APP6. “-,” absence, and “+,” presence.

^aThe rsmA and estT genes detected with a loose criteria with the CARD database: 60% identity with 60% coverage.

^b & cThe floR gene present in the plasmid pAp_ZJNH (pMAF6-like, GenBank: CP100663.1)^b, or pAp-AH and pAp_ZJXS (pIV86-type, GenBank: OQ325044.1)^c.

The floR gene is present in all three strains of our study (Table 5) and matches with their florfenicol resistance (Table 2). Initially, only the strain ZJXS2022 was found to contain the pIV86-type plasmid carrying floR from the sequencing data (pAp_ZJXS, Fig. S6 left) at high nucleotide similarity (99.85% at 100% coverage) to pIV86 from Actinobacillus indolicus strain IV86 (GenBank: OQ325044) (Tab. S7). For the strains AH2022 and ZJNH2023, the contigs from the original sequencing data were retrieved for analysis by PlasmidSPAdes with MOB-Suite, followed by BLAST. We found that the strain AH2022 also harbors the pIV86-type plasmid at 99.96% similarity to pIV86 (pAp_AH, Fig. S6 middle). We verified the correct amplicons of the pIV86-type plasmids in ZJXS2022 and AH2022, but not in ZJNH2023, by using relevant primer pairs (Figs. S1 & S7). Nevertheless, BLAST search showed that the putative plasmid in ZJNH2023 has the best hit for the floR-containing pMAF6-type plasmid (CP100663, in a Chinese App strain) at 99.76% similarity and 100% coverage (Tab. S7), but with organizational differences in four fragments. The plasmid pAp_ZJNH in the strain ZJNH2023 is identified as pMAF6-type upon re-sequencing of the plasmid extract (Fig. S6 right) and PCR verification (Fig. S7).

Discussion

Comparative genomic analysis of microbial pathogens could provide insights into the genetic virulence features governing the pathogenicity of different strains of the same species or among the strains of different serovars as well as genetic diversity and evolutional process. Since the completion of sequencing of the first few App genomes in 2008 [12] and 2010 [16], a total of 192 genomes of App strains have been deposited in the NCBI database with 11 from Chinese strains (www.ncbi.nlm.nih.gov/datasets/genome/?taxon=715), as accessed in 2 September 2025). However, there have been few reports of comparative genomic studies of this pathogen until recently [11,17–19]. Here, we report the genomic features of three recent App strains isolated from diseased pigs in several swine farms in southeastern Chinese provinces Zhejiang and Anhui. Although App genomes are generally conserved, there are insertions of large GIs and prophages of different numbers in our strains, leading to more CDSs (65 to 193 more) than the average of 2076 in the sequenced genomes (Table 3). We have found some distinct features of these new strains in terms of virulence genes, prophages, GIs, plasmids, and ARGs.

While all serovars of App strains are pathogenic, the severity of infection could differ significantly from high morbidity and mortality to asymptotic carriage, depending in part on the production of the RTA toxins ApxI to ApXIV [2,46]. The App strains producing ApxI and ApxII toxins are considered the most virulent [2,46]. However, the adhesins are equally important in pathogenesis by initiation of bacterial colonization and persistence on the mucosal surface. Besides fimbrial proteins ApfA and Tfp responsible for App adhesion [47,48], there are also the type IVb pili formed by the flp locus genes [46,49] as well as trimeric autotransporter adhesins Apa1 and Apa2 [45]. Of the three strains in this study, the major differences of virulence-related genes are those expressing cytotoxins (ApxIs and ApxIIIs) and adhesins (Apa1 and ApaI), and those in the flp locus related to the formation of type IVb pili (Table 4). Higher pathogenicity of the serovar 5 strain ZJNH2023 might be due to combined roles of the full set of ApxI genes as well as Apa1/Apa2 and intact flp locus genes which are absent or incomplete in serovar 8 and 15 strains AH2022 and ZJXS2022.

The App strains possess about 1700 core genes and have genomic plasticity because of horizontal gene transfer (HGT) [16,43]. HGT can occur via transposition in the form of transposons or by transduction/integration of prophages. App prophages have been reported in recent studies [17,43]. Of the 35 types of prophages found in this study, the majority are partial or incomplete and few were intact, including the three Mannheimia phages present in different serovar strains. The three strains of this study have 3–5 more prophages than those of other studies (Tab. S4) and the strain ZJXS2022 contains intact Mannheimia phage and phage Mu in two different GIs (Tabs. S4 & S5, Fig. S5). There are distinct GIs in our strains, as compared with those of corresponding serovars of other studies: serovar5_GI-1 (only in ZJNH2023) and serovar5_GI-4 (in ZJNH2023 and the other Chinese serovar 5 strain APP6), serovar8_GI-1 (AH2022), as well as serovar15_GI-2, _GI-3, and _GI-5 (ZJXS2022) (Tab. S5). Different types of transposases, including ISApl1, ISEc29, ISShfr9, IS3000, and ISVsa5, are present in these GIs (ranging from 3 to 5 per GI). Besides, there were integrases (IntA and IntS) in the distinct GI-1 and GI-4 of serovar5 strain ZJNH2023. These findings suggest that transposons and prophages are involved in HGT in these App strains.

There are more ARGs (4 to 9 ARGs) in our strains than the other strains of respective serovars (0 to 2 ARGs) (Figure 1, Table 5). Different from the reported presence of ARGs in plasmids [17,43], we found that ARGs, such as tet(B), dfrA6, sul2, and those resistant to aminoglycosides, were present in the GIs of strains AH2022 and ZJXS2022, indicating that GI-mediated HGT is involved in ARG transfer and dissemination in the App strains on the swine farms in this region probably as a result of imprudent use of antimicrobials in the past. A recent report indicates that there was considerable resistance of App isolates from pigs in Zhejiang and surrounding areas (the same region as in our study) to the same antimicrobials used (11%–24%) [50]. We found that the three strains showed multiple resistance to 5–6 antimicrobials: florfenicol, chloramphenicol, tetracycline, and kanamycin via the floR, tet(B), and aph(3’)-Ia genes (Table 5). The strain AH2022 was resistant to tilmicosin via the estT gene expressing macrolide hydrolase EstT. However, no known resistant genes, including estT, were found for the tilmicosin-resistant strain ZJNH2023. This might be due to mutations in the 20S rRNA as reported for other swine pathogens, Pasteurella multocida [51] and Haemophilus parasuis [52].

Because of the presence of the floR gene in the three strains and their resistance to florfenicol from initial sequencing data, we speculated that the other two strains AH2022 and ZJNH2023 might also have the floR-containing pIV86 plasmid, similar to the strain ZJXS2022. With sets of the primers targeting the overlapping fragments of the circular DNA according to pIV86, we found that the band profiles of the strain AH2022, but not ZJNH2023, were similar to ZJXS2022 (Fig. S7). Retrieval of the sequenced contigs confirmed the presence of the pIV86-type plasmid in AH2022 (Fig. S6), and also suggested a pMAF6-type plasmid in ZJNH2023 which is verified by re-sequencing as pMAF6-like with organizational differences to pMAF6 (Fig. S6). There are several floR-containing plasmids in the App strains not only from China, but also from other countries such as Italy where the pAp-floR plasmid was shown to have high nucleotide identity with the pMVSCS1 plasmid from Mannheimia varigena MVSCS1 of porcine origin [53]. This, together with the presence of the pIV86-type and pMAF6-type plasmids in the Chinese strains of bacterial species of swine origin other than A. pleuropneumoniae (Tab. S7), suggests transfer of the floR-carrying plasmids between App strains or among different bacterial species in the pig farming environments.

In conclusion, our comparative genomic analysis has revealed that the serovar 5 strain ZJNH2023 carries the cytotoxic ApxI gene set and those involved in adhesion that might have contributed to its higher pathogenicity than the other two strains. There are more prophages, GIs and ARGs in the three strains than those of the same serovars from other studies. Our findings indicate that distinct GIs and plasmids might have involved in horizontal transfer of ARGs in the App strains circulating on the pig farms in China. Further expanding of similar work involving more clinical isolates in diverse geographical regions would enhance our understanding of the genomic epidemiology of this particular swine pathogen and the extent of ARGs dissemination on the swine farms. Knowledge of predominant virulent serovars would help facilitate development of more effective vaccines.

Funding Statement

This research was supported by the ‘Pioneer’ and ‘Leading Goose’ R&D Program of Zhejiang (No.2025C01138, 2023C02023, 2023C02047), Agricultural Science and Technology Cooperation Program of Zhejiang Province (No. 2023SNJF049).

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The genome sequences of all Actinobacillus pleuropneumoniae strains were uploaded to GenBank with accession number CP143255 (ZJNH2023), CP141949 (AH2022), and CP141951 (ZJXS2022). The accession numbers of plasmid sequences include CP141952 for pAp_ZJXS (strain ZJXS2022), CP199936 for pAp_AH (strain AH2022) and CP199937 for pAp-ZJNH (strain ZJNH2023).

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/21505594.2026.2620200.