CARBON SOURCE UTILISATION AND EVALUATION OF THE BIOLOG SYSTEM IN THE IDENTIFICATION OF ACTINOBACILLUS PLEUROPNEUMONIAE

Sixty-eight Actinobacillus pleuropneumoniae strains were isolated from porcine acute pleuropneumonia cases from different parts of Hungary between 2000 and 2014. A total of 41 isolates were identified as A. pleuropneumoniae biotype I and 27 strains as biotype II based on cultural, morphological and biochemical characteristics. The aim of this study was to evaluate metabolic fingerprinting in the species-level identification of A. pleuropneumoniae isolates. Utilisation of carbon sources by these field isolates and six reference strains was characterised by the Biolog system (GN2 Microplate, MicroLog3 Version 4.20.05 software). Twenty-nine field strains were correctly identified by the Biolog system as A. pleuropneumoniae , 36 strains as A. lignieresii , two strains as H. paraphrohaemolyticus and one strain as A. equuli after 24 h of incubation. Among the six A. pleuropneumoniae reference strains the Biolog system identified one strain as A. pleuropneumoniae , four as A. lignieresii and one as H. paraphrohaemolyticus . There was no correlation between biotypes and serotypes of A. pleuropneumoniae and the carbon source utilisation pattern and species identification by the Biolog system. Our data indicate that the efficacy of the Biolog system used here could be improved by including phenotypes of more A. pleuropneumoniae strains representing a wider geographical occurrence into the database.

All 18 serovar reference strains are able to express some of the four different Apx toxins belonging to the pore-forming repeats-in-toxin (RTX) group (Shin et al., 2011;Sárközi et al., 2015;Bossé et al., 2018). Three toxins, ApxI, ApxII and ApxIII are responsible for haemolysis and cytotoxic damage of the lung cells (Sthitmatee et al., 2003), but these toxins are produced by other bacterium species, too, not only by A. pleuropneumoniae (Schaller et al., 2001). The ApxIV toxin gene is species specific, it can be found only in A. pleuropneumonae strains, and ApxIV toxin is produced only in live animals (Schaller et al., 2001). Although A. lignieresii is the species most closely related to A. pleuropneumoniae as determined by DNA-DNA hybridisation and comparison of the 16S rRNA sequences (Borr et al., 1991), there is no apxIVA gene in A. lignieresii (Schaller et al., 2001).
There are six members of the Actinobacillus genus which are nowadays recognised as significant causes of diseases in animals: A. pleuropneumoniae, A. suis, A. lignieresii, A. equuli, A. seminis, and A. capsulatus (Rycroft and Garside, 2000). Many species of the Actinobacillus genus other than A. pleuropneumoniae can be found in tonsils of swine, such as A. minor, A. porcinus, Bisgaard's Taxon 10, A. rossii, and A. porcitonsillarum (Lowe et al., 2011;. Actinobacillus muris, A. hominis and A. ureae are species of little veterinary impact (Christensen and Bisgaard, 2004).
Actinobacillus lignieresii can be found on the mucous membranes of cattle and is able to cause wooden tongue or lesions in the oral cavity and the regional lymph nodes . Differentiation of A. pleuropneumoniae biotype II strains and A. lignieresii is difficult because of their close phylogenetic relationship and common characteristics (Rycroft and Garside, 2000).
There are different methods for the identification of bacteria. In addition to the detection of genus-, species-and serotype-specific genes, identification based on phenotypic characteristics is also widely used. Besides the traditional identification using cultural, morphological, biochemical and serological features (Barrow and Feltham, 2003), several identification systems based on the examination of phenotypic characteristics are available on the market. Different manual biochemical identification systems like API The Biolog Microbial Identification Systems (Biolog Inc., Hayward, CA, USA) are available in manual, semi-automated and fully automated forms; they identify the bacteria on the basis of utilisation of carbon sources (Wong et al., 1992). Their databases include A. pleuropneumoniae, A. lignieresii, A. hominis, A. equuli and A. suis from the Actinobacillus genus.
The aim of this study was to examine metabolic fingerprinting of field isolates and reference A. pleuropneumoniae strains based on the utilisation of 95 carbon sources and to evaluate this method in the identification of A. pleuropneumoniae strains.

Bacterium strains
Sixty-eight A. pleuropneumoniae field isolates were included in the examination; all were isolated from lung samples collected in slaughterhouses and from postmortem cases of acute porcine pleuropneumonia, submitted to our laboratory from different Hungarian swine farms between 2000 and 2010. One of them was suggested as reference strain of serotype 16 isolated in 2014. Six serotype reference strains of A. pleuropneumoniae K17 (serotype 5a), L20 (serotype 5b), CVI13261 (serotype 9), D13039 (serotype 10), 56153 (serotype 11) and N273 (serotype 13) provided by Dr. Ø. Angen (Danish Veterinary Laboratory, Copenhagen) were included in the examinations.
The A. pleuropneumoniae strains were isolated on Tryptone Soya Agar (TSA, Biolab Ltd., Hungary) cross-inoculated with Staphylococcus aureus, and subcultured on chocolate agar with added 50 μg/ml NAD (Biolab Ltd., Hungary), both containing 10% defibrinated sheep blood. Cultures were incubated at 37 °C for 24 h in aerobic environment with the addition of 5% carbon dioxide. They were identified using standard methods (Barrow and Feltham, 2003), and most strains were serotyped . After identification, the isolated A. pleuropneumoniae strains were stored at -80 °C until further examination.

Carbon source utilisation
A 96-well automated MicroLog MicroStation System with GN2 Microplates (Biolog Inc., Hayward, CA, USA) was used for the characterisation of carbon source utilisation. Microplates were set up and analysed following the manufacturer's instructions with minor modification. Single colonies of biotype I of A. pleuropneumoniae were subcultured three times on Biolog Universal Growth (BUG) agar plates with NAD, and biotype II strains were cultured on BUG agar containing 10% defibrinated blood. Two pure colonies from the third subculture of each strain were inoculated on two chocolate plates with NAD or blood agar plates evenly covering the whole surface of the plate. Plates were cultured at 37 °C and 5% CO 2 . After 24-h incubation, the thin and confluent layer of A. pleuropneumoniae was collected with a cotton swab and suspended in 18 ml inoculation fluid (GN/GP IF) to obtain a homogeneous suspension. The turbidity of the bacterium suspensions was set to 20 ± 2% using the Biolog Turbidimeter. GN2 MicroPlates were inoculated with 150 µl bacterium suspension per well and incubated at 37 °C in 5% CO 2 atmosphere. Metabolic activity was determined by visual reading of the plates after 24 h. The results were evaluated and a dendrogram showing the metabolic relationships between the strains was produced by Biolog MicroLog3 software (Biolog Inc., USA, Version 4.20.05).

Identification of bacterium isolates
All 68 field isolates were Gram-negative, < 2 µm, coccoid rods. They produced small grey colonies surrounded by a narrow β-haemolytic zone. They all produced urease, they were oxidase positive but catalase negative. All strains proved to be A. pleuropneumoniae, 41 strains needed NAD, and 27 strains were able to grow without NAD.

Carbon source utilisation
The carbon source utilisation of the A. pleuropneumoniae strains is presented in Table 1. There were no major differences between biotype I and II strains of A. pleuropneumoniae in carbon source utilisation pattern. All strains were able to metabolise 20 carbon sources and 1-99% of the strains could utilise 27 further carbon sources after 24-hour-long incubation. Twenty-nine out of the 68 field isolates were identified by the Biolog system as A. pleuropneumoniae, 36 strains as A. lignieresii, two strains as H. paraphrohaemolyticus, and one strain as A. equuli.
The dendrograms show the similarity of all the A. pleuropneumoniae strains and that of biotype I and II ones (Figs 1, 2 and 3).

Discussion
In addition to widely used nucleic acid typing methods, identification systems based on the detection of various phenotypic characteristics are also available on the market and used both in human and veterinary medicine. Biolog Microbial Identification Systems are used successfully for the identification of Gram-positive and Gram-negative bacteria (Gyuranecz et al., 2010;Zasada and Mosiej, 2018).
Our results show that identification of the primary pig pathogen A. pleuropneumoniae based on carbon source utilisation using the Biolog system has only limited value due to the high similarity of A. pleuropneumoniae and A. lignieresii. If the metabolic fingerprint shows questionable results, these two bacterium species can be appropriately differentiated by taking into consideration the pathological origin of the bacterial isolate (A. pleuropneumoniae: haemorrhagic, necrotic fibrinous pleuropneumonia of swine, A. lignieresii: granulomatous mastitis of pigs, or tongue, lymph node, ruminal wall or skin lesions of ruminants) and some cultural (growth on MacConkey agar of A. lignieresii) and haemolytic features [haemolysis on blood agar and positive CAMP test with Staphylococcus aureus (A. pp.)] of the isolates.
There was no correlation between the biotypes and serotypes of A. pleuropneumoniae and carbon source utilisation pattern and species identification by the Biolog system.
Comparing our results with the Biolog standard of A. pleuropneumoniae and A. lignieresii, it is evident that some patterns of A. pleuropneumoniae strains included in the Biolog database have not been represented in our study, as certain carbon sources were not utilised at all by our isolates. No major difference could be seen between the carbon source utilisation of biotype I and II strains of the field isolates.
The dendrograms based on carbon source utilisation show a high level of similarity, especially in the case of biotype II strains of A. pleuropneumoniae, where the difference was below 5%. A higher variability was seen in the case of biotype I strains, but the difference was below 7.5% in this case as well. Our data confirm the results of other authors on the low variability of A. pleuropneumoniae strains (Fussing et al., 1998;Kokotovic and Angen, 2007;Sassu et al., 2018;Ito et al., 2018).
Actinobacillus pleuropneumoniae strains show a high level of antigenic variability in different geographical locations Perry et al., 2012), and a similar metabolic variability could be expected in the utilisation of carbon sources, too. The efficacy of the Biolog system could be improved by including phenotypes of more A. pleuropneumoniae strains representing a wider geographical occurrence into the database.