Characterisation of Pasteurella multocida isolates from pigs with pneumonia in Korea

Background Pasteurella multocida is responsible for significant economic losses in pigs worldwide. In clinically diseased pigs, most P. multocida isolates are characterised as subspecies multocida, biovar 2 or 3 and capsular type A or D; however, there is little information regarding subspecies, biovars, and other capsular types of P. multocida isolates in Korea. Here, we provided information covering an extended time period regarding P. multocida in pigs with pneumonia in Korea using phenotypic and genotypic characterisations and data associated with the minimum inhibitory concentrations. Results The overall prevalence of P. multocida between 2008 and 2016 was 16.8% (240/1430), with 85% of the P. multocida isolates (204/240) coinfected with other respiratory pathogens. Of the 240 isolates, 166 were included in this study; all of these P. multocida isolates were characterised as subspecies multocida and the most prevalent phenotypes were represented by biovar 3 (68.7%; n = 114) and capsular type A (69.9%; n = 116). Additionally, three capsular type F isolates were identified, with this representing the first report of such isolates in Korea. All biovar 1 and 2 isolates were capsular types F and A, respectively. The virulence-associated gene distribution was variable; all capsular type A and D isolates harboured pmHAS and hsf-1, respectively (P < 0.001), with type F (biovar 1) significantly correlated with hsf-1 (P < 0.05) and pfhA (P < 0.01), biovar 2 highly associated with pfhA and pmHAS, and biovar 3 significantly correlated with hsf-1, pmHAS, and hgbB (P < 0.001), whereas biovar 13 was related only to hgbB (P < 0.05). The highest resistance rate was found to be to oxytetracycline (63.3%), followed by florfenicol (16.3%). Conclusions P. multocida subspecies multocida, biovar 3, and capsular type A was the most prevalent isolate in this study, and our findings indicated the emergence of capsular type F in Korea. Moreover, prudent use of oxytetracycline and florfenicol is required because of the identified high resistance rates. Further studies are required for continuous monitoring of the antimicrobial resistance, prevalence, and epidemiological characterisation of P. multocida, and experimental infection models are needed to define the pathogenicity of capsular type F.


Background
Pasteurella multocida (P. multocida) is a commensal and opportunistic pathogen of the oral, nasopharyngeal, and upper respiratory tract [1] and the causative agent of a wide range of infections leading to high economic impact [2]. In pigs, P. multocida is associated with progressive atrophic rhinitis (PAR), and together with other respiratory pathogens, plays a significant role in porcine respiratory disease complex (PRDC) [3][4][5][6]. P. multocida prevalence has been reported as 8.0% in diseased pigs with pneumonia or PAR in China, and from 10.3 to 15.6% in pigs with pneumonia in Korea. Additionally, P. multocida constitutes 15.6% of isolated respiratory pathogens in the United States [3,5,7,8].
Antimicrobial resistance in pathogenic bacteria from food animals and the environment has become a global public health issue. Although beta-lactams, trimethoprim combination, florfenicol, macrolides, and tetracyclines have been shown to be the best antimicrobials for treating PRDC [6], resistance to these antimicrobials has been detected previously in P. multocida in many countries [3,[22][23][24]. In Korea, P. multocida isolates from pigs are sensitive to most antimicrobial agents, including ampicillin, ceftiofur, tilmicosin, and enrofloxacin, other than tiamulin [7].
To the best of our knowledge, only short-term studies have been performed to characterise porcine P. multocida isolates in Korea. This long-term study was carried out to provide baseline information regarding a large collection of P. multocida isolates from clinically diseased pigs by determining the distribution and association of capsular types, biovars, extensive virulence-associated gene profiles, and antimicrobial-resistance patterns.

Subspecies, biovar, and capsular type determination
The distribution of biovars and capsular types among the studied P. multocida isolates is shown in Table 3. All 166 isolates were identified as P. multocida subspecies multocida, which produces acid from sorbitol and glucose but not from dulcitol, lactose, and maltose. Most ODC-producing isolates belonged to biovar 3 (68.7%), followed by biovars 2 (21.1%) and 1 (1.8%). Interestingly, 14 isolates (8.4%) displayed identical carbohydrate fermentation results to biovar 3, except for ODC activity, and were thus assigned to biovar 13. All biovar 1 and 2 isolates comprised capsular type F and A, respectively (P < 0.001), whereas biovar 3 isolates comprised capsular types A and D (P < 0.001), and biovar 13 comprised capsular types A and D (P > 0.05). Capsular type A (69.9%) isolates were the most prevalent, followed by types D (28.3%) and F (1.8%), with none of the isolates in this study identified as type B or E. Importantly, this is the first report of capsular type F/biovar 1 isolation since 2014 ( Table 3).

Discussion
Our findings showed that P. multocida isolates were prevalent (16.8%) in pig farms and the second most frequently isolated bacterial pathogen from diseased pigs, following S. suis (17.6%). This was consistent with previous studies in Korea that reported the prevalence of P.   multocida to be between 10 and 15.6% [7,8]. The infections in this study comprised of a mix of P. multocida (85.0%) with other respiratory pathogens, particularly PRRSV (61.3%; P = 0.0001). Therefore, veterinary practitioners and surveillance stakeholders should consider coinfection with various pathogens that might exist in a given herd for PRDC control. We characterised 166 P. multocida isolates by determining their subspecies, biovar, capsular type, virulence-associated genes, and MIC. To the best of our knowledge, this is the first report of biovar prevalence in Korea. All isolates belonged to subspecies multocida, and the most prevalent type was biovar 3 (68.7%), which is consistent with the results of previous studies of P. multocida recovered from pigs [1,10,13]. P. multocida biovar 1 is frequently isolated from poultry, but not pigs [1]. We found that the prevalence of biovar 13 was 8.4%, which is slightly higher than that in other countries, such as Australia (2.0%) and Hungary (4.8%) [10,11]. In agreement with numerous previous studies, the dominant P. multocida capsular types recovered from pneumonic pig lungs were capsular types A (69.9%) and D (28.3%) [1,15,16,26]. Additionally, capsular type B is the etiological cause of septicaemic pasteurellosis, whereas type F is rarely reported in pigs [1,14]. Interestingly, capsular type F has been isolated in Korea post 2014, although at relatively low proportions (n = 3; 1.8%), the prevalence of which is consistent with that reported in other European studies [0.3% (Germany), 1.0% (UK), and 2.4% (Spain)] [1,2,16]. A recent Chinese experimental study indicated that pig-origin capsular type F isolates are associated with porcine pneumonia and exhibit high pathogenicity in pigs [27]. Additionally, we found that P. multocida capsular type F was the only relevant respiratory pathogen isolated from three growing pigs with moderate-to-severe suppurative bronchopneumonia with fibrous/fibrinous    [25] pleuritis. This represents the first report identifying capsular type F isolates in Korea; therefore, the pathogenic significance of type F in pigs needs to be elucidated. Virulence genotyping is a useful typing method for molecular characterisation of bacterial pathogens and has been previously applied to P. multocida [1,3]. Although oma87, ompH, plpB, psl, fimA, hsf-2, sodA, sodC, exbB, ExbBD-tonB, fur, hgbA, nanB, nanH, and ptfA were uniformly distributed among the isolates tested, none possessed tbpA, which agreed with the results of previous pig studies [1][2][3]17]. The wide distribution of these genes indicates their importance for the survival of P. multocida within the host environment. Additionally, the virulence factors involved in cross-protection might constitute potential vaccine candidates, regardless of capsular type [3]. However, previous studies demonstrated that several non-uniformly distributed virulence-associated genes exhibit significant relatedness with specific capsular types [1,3,8,17]. As shown in Table 4, all capsular type A and D isolates harboured pmHAS and hsf-1, respectively, and most type D isolates harboured hgbB (P < 0.001). In this study, capsular type F displayed virulence-associated gene profiles similar to those of capsular type D (hsf-1 + hgbB + ), except for pfhA. Previous studies reported toxA as clearly associated with type D [1,3,7,17]; however, we found that only one of the 47 type D (2.1%) isolates and 6.9% of type A isolates harboured toxA. These results, however, are not significant because most of the isolates were from pneumonic lesions and not from turbinates with PAR. Similar to a previous report, distinct associations were observed between the virulence-associated gene profiles of toxA, hgbB, and pfhA and biovars, except for biovar 13 [1]. All biovar 1, 2, and 13 isolates exhibited toxA − hgbB + pfhA + (P < 0.001), tox-A − hgbB − pfhA + (P < 0.001), and toxA − hgbB + pfhA − (P < 0.05) profiles, respectively, and most biovar 3 isolates displayed a toxA − hgbB + pfhA − profile (P < 0.001). Additionally, toxA was found only in biovar 3 isolates (toxA + hgbB + pfhA − ; P < 0.05).
Swine diseases have become co-infected with immunosuppressive diseases, leading to antimicrobial treatment failure and frequent resistance occurrence. Treatment against P. multocida infections commonly includes broadspectrum antimicrobials [3]. In this study, beta-lactams (penicillin, ampicillin, and ceftiofur), macrolides (tulathromycin and tilmicosin), and fluoroquinolone (enrofloxacin) were found to be more effective than oxytetracycline and florfenicol. Therefore, these agents are recommended as empirical antimicrobials for the treatment of P. multocida infection. Tetracycline resistance has previously been reported in P. multocida isolates worldwide [3,6,25,28,29]. Its prevalence in the present study was found to be 63.3%, which is similar to the prevalence in China (58.0%) and North America (53.4%) [3,28] but higher than that in Australia (28.0%) and European countries (20.4%).
Previous studies recommended the use of florfenicol for the treatment of infections caused by P. multocida, because florfenicol-resistance rates are very low (0-2%) in China, North America, Australia, and Europe [6,25]; however, the present study showed a relatively higher resistance (16.3%). According to the Korea Animal Health Products Association, tetracyclines and florfenicol are the most commonly used antibiotics in Korean pig husbandry [30], with their frequent use reflected in the resistance rates in the present study. Based on the occurrence of high rates of tetracycline and florfenicol resistance, these antimicrobial agents should be used carefully and accompanied by susceptibility tests. Additionally, continuous surveillance of antimicrobial resistance in respiratory pathogens, including P. multocida, is required due to the increasing use of therapeutic antimicrobials and emergence of new resistant strains.
This study was conducted to determine the phenotypic and genotypic characteristics of swine P. multocida isolates in Korea. However, the collected samples cannot be representative of current P. multocida isolates in Korean swine farms, given that the number of isolates submitted annually varies, and the isolates used in this study originated from diagnostic samples with unknown antimicrobial-treatment history. However, a large-scale study for the characterisation of clinical lung samples of P. multocida isolates would sufficiently broaden the understanding of P. multocida as a respiratory pathogen.

Conclusions
This represents a comprehensive report of P. multocida isolates in pigs in Korea. Our findings provide scientific information for further research, including development of vaccine candidates and guidelines for antimicrobial use in veterinary medicine. Moreover, the low discriminatory power of phenotypic characterisation limits the scope of adequate epidemiological information; therefore, different genotyping techniques using pulsed-field gel electrophoresis or multilocus sequence typing might be required to further elucidate the epidemiology of P. multocida and its genetic relatedness.

Bacterial isolation and identification
In total, 1430 lung samples were collected from pigs (suckling pigs, 9%; weaning pigs, 49%; growing-finishing pigs, 23%; and unknown, 19%) with pneumonic gross lesions from 514 farms nationwide between 2008 and 2016. All lung samples were submitted to the Animal and Plant Quarantine Agency for differential diagnosis of porcine respiratory diseases, including APP, HPS, S. suis, T. pyogenes, MHP, MHR, PRRSV, PCV2, and SIV. Following gross and histopathologic examination, samples were cultured on 5% sheep blood agar, chocolate agar (Asan Pharm. Co., Ltd., Seoul, Korea), and MacConkey agar (Becton Dickinson, Sparks, MD, USA) and then incubated aerobically at 37°C for 48 h. Suspected mucoid and non-haemolytic colonies were subjected to Gram staining and biochemical identification using the VITEK II system (BioMérieux, Marcy l'Etoile, France). Identification was further confirmed by species-specific PCR assay for amplification of kmt1 (Table 1) [19]. All P. multocida isolates were stored at − 80°C until use to determine the subspecies, biovar, and capsular type. Previously reported methods were used to differentiate between P. multocida and other pathogens [3,31,32].

Subspecies and biovar determination
The confirmed P. multocida isolates were classified into three subspecies (multocida, septica, and gallicida) based on sorbitol and dulcitol fermentation [9]. Additionally, isolates were assigned to one of the established biovars based on their ability to ferment carbohydrates (sorbitol, dulcitol, maltose, xylose, glucose, trehalose, lactose, and arabinose) and produce the ODC enzyme [10].

Antimicrobial-susceptibility testing
The MIC of all isolates (n = 166) was determined using the standard broth microdilution method with the Sensititre system (TREK Diagnostic System; Thermo Fisher Scientific, Cleveland, OH, USA) and commercially prepared 96-well antimicrobial testing plates containing 18 different agents (BOPO6F; TREK Diagnostic Systems). The following antimicrobials were tested: penicillin, ampicillin, ceftiofur, florfenicol, gentamicin, neomycin, chlortetracycline, oxytetracycline, clindamycin, enrofloxacin, danofloxacin, trimethoprim/sulfamethoxazole, sulfadimethoxine, spectinomycin, tulathromycin, tylosin tartrate, tilmicosin, and tiamulin. Escherichia coli ATCC 25922 was tested for quality control purposes. As shown in Table 6, the MICs were interpreted according to CLSI guidelines for oxytetracycline, florfenicol, penicillin, ampicillin, enrofloxacin, tulathromycin, ceftiofur, and tilmicosin or those of a previous study describing analysis of trimethoprim/sulfamethoxazole, for which CLSI breakpoints were not available [25,33]. The overall MIC 50 and MIC 90 values (i.e., the lowest concentrations at which growth was inhibited by 50 and 90%, respectively) for each antimicrobial were determined for all isolates.