Serovar and Virulence Genes of Glaesserella(Haemophilus) Parasuis Isolates from the Nasal Cavity of Live Piglets

Glaesserella (Haemophilus) parasuis (G. parasuis) is a commensal bacterium in the swine upper respiratory tract that can cause Glässer’s disease, particularly in piglets. In this study, we detected the serovars and 19 known virulence genes (VGs) of H. parasuis isolates from the nasal cavities of live piglets from the south of China.Serovar 10 (17.9%) was the most prevalent, followed by serovars 15 (14.5%), 6 (12.0%), 8 (11.1%), 4 (8.5%), 7 (7.7%), 9 (7.7%), 1 (7%), 5/12 (4.3%), and 2 (0.9%). This differs from previous studies on common G. parasuis serovars. The detection rate of 19 VGs ranged from 1.7% to 95.2%, with vacJ and clpP (95.7%) as the most prevalent. The G. parasuis isolates belonging to the same sequence type and serovar harbored different VGs, and all isolates exhibited considerable genetic heterogeneity. Signicant correlations were found between VGs and serovars, different pathogenic serovar groups, and members of clade 2 (based on MLST). To our knowledge, this is the rst research to examine the characteristics of G. parasuis nasal cavity isolates from live piglets in the south of China. The results complement epidemiological data of G. parasuis and will help the scientic community understand the extreme genetic diversity and pathogenesis of G. parasuis, which will aid in the development of G. parasuis vaccines.

To date, 15 serovars have been identi ed, in addition to some non-typable (NT) strains (Kielstein P and Rapp-Gabrielson,1992;Jin et al., 2006 ). Serovar identi cation of the isolates is the main basis for designing vaccination programs . Some earlier studies suggested that G. parasuis serovars were a virulence marker and could be divided into three pathogenic groups (Oliveira and Pijoan, 2004). However, later studies found that isolates allocated into non-pathogenic serovars can also cause disease, and virulence of the isolates allocated to the same serovar can vary greatly ( Comparing the pathogenicity difference between wild-type strains and gene knockout mutants is an important method for evaluating the role of putative VGs in pathogenicity. In addition, molecular epidemiology and bioinformatics studies are often used to discover and evaluate putative VGs, and some VGs have already been evaluated by these methods. It is generally believed that a single VG may not be a decisive factor in triggering the pathogenesis of multifactorial diseases such as Glässer's disease, and the pathogenesis of bacteria often depends on the interaction and expression regulation of many VGs.
Thus, a comprehensive analysis of VGs in clinical isolates may be helpful to predict the pathogenicity of novel G. parasuis isolates as they are identi ed.
Although the characteristics of G. parasuis isolates from clinical cases have been extensively studied, an in-depth analysis of G. parasuis isolates from the swine upper respiratory tract has not been performed. In this study, we analyzed the characteristics, including serovars and VGs, of G. parasuis isolates from the nasal cavities of live piglets in three provinces in the south of China. Our results provide more information on the epidemiology and pathogenesis of G. parasuis.

Materials And Methods
Identi cation and serotyping One hundred seventeen G. parasuis isolates were isolated from the nasal cavities of live piglets without obvious clinical symptoms of Glässer's disease between 2007 and 2016 in three provinces (Guangdong, Jiangxi, and Shanghai) in the south of China. The G. parasuis strains were identi ed by NAD-dependency and 16S rRNA PCR (Angen et al., 2007). The isolates underwent molecular serotyping via a multiplex PCR assay described in Howell et al. (2015).

VG analysis
Nineteen VGs were analyzed using PCR as previously described details of all primers used are listed in Table 1. MLST A MLST analysis was carried out using a method previously described (Olvera et al., 2006;Mullins et al., 2013). A neighbor-joining tree was built using the MEGA version 5.0 software based on the MLST target sequences.

Association between MLST and VGs
The MLST analysis revealed two major clades (clade 1 and clade 2) based on the MLST target sequences of 43 G. parasuis isolates. Clade 1 includes 37 isolates of serovars 1, 2, 4, 6, 7, 8, 9, 10, 15, and NT, harboring 1 to 11 VGs each. Clade 2 includes 6 isolates of serovars 1, 4 and 5, harboring 8 to 16 VGs each (Fig. 4). Interestingly, isolates in the second clade had a signi cantly increased probability of containing the VGs vta1, nhaC, hhdA, hhdB, lsgB, H0254, wbgY, mB, and 1373 (p< 0.05, Table 4). Those authors did not identify any isolates representing serovars 14 and 15. In the current study, we did not isolate any G. parasuis strains from serovars 3 and 11, and we only isolated a single strain from serovar 2. This suggests that serovars of G. parasuis from the swine nasal cavity exhibit a complex regional distribution across provinces in China.
In both the current study and the study conducted by Zhang et al. (2019), the detection frequency of serovars 4 and 5/12 was relatively low. Strains in serovars 4 and 5 are widely regarded as pathogenic strains, and they are most often identi ed from pigs with Glässer's disease. Although the detection frequency of serovars 4 and 5/12 was not high in live piglets, these isolates may nonetheless cause disease when an animal is under stress. Of note, the dominant serovars identi ed in this study, serovar 10 and serovar 15, were previously considered to be highly and moderately pathogenic, respectively. These two serovars have rarely been isolated in diseased pigs in China. Further attention and research are required to determine whether the presence of strains from serovars 10 and 15 in the respiratory tract of live piglets would cause localized disease, or even a potential disease epidemic.
In this study, all G. parasuis isolates were divided into four clusters according to the presence of VGs. Though serovars 2, 5, 8, 9, 10, and 15 were only distributed in one cluster, isolates belonging to the same serovar harbored different VGs. These differences were also present among strains that belonged to the same ST and serovar. For example, strains SG25 and N1-24, isolated from different farms, were both allocated to ST185 and serovar 8, and possessed seven identical VGs. However, strain SG25 had ve more VGs than N1-24. Similarly, strains OY2 and QY6-1, isolated from the same farm, were allocated to ST255 and serovar 15, but strain QY6-1 has one more VG (rfaE) than OY2. Interestingly, strain QY6, isolated from the nasal cavity of the same piglet as strain QY6-1, also harbored rfaE. These results suggest that G. parasuis isolates may undergo multiple gene exchanges while coexisting in the respiratory tract. The VGs of isolates allocated to the same ST and serovar varied greatly, which may lead to differences in the pathogenicity and immunogenicity of strains belonging to the same ST and serovar. Once these strains invade the host tissues and organs, they may cause localized disease and eventually become epidemics. At that point, even if the serovars of commercially available vaccines and pathogenic strains were the same, the differences in VGs may lead to immune failures. That scenario would pose a substantial challenge to the development of a new vaccine. Van 2018) reported that the detection frequency of hhdA and hhdB was 36% and 13.3%, respectively, which differs from our results for most of the above VGs. Although previous studies have shown that the VGs lsgB, fhuA, capD, HPM-1372, and HPM-1373 were not observed in any isolates from non-pathogenic serovar group, our results showed that 8 of 43 isolates from the non-pathogenic serovar group were positive for lsgB, 16 were positive for fhuA, 39 were positive for capD, 8 were positive for HPM-1372, and 1 was positive for HPM-1373. Our results indicate that the distribution of VGs in G. parasuis is diverse and complex. Olvera et al. (2012) reported that isolates without vtaA1 are generally avirulent. In this study, the presence of vta1 was associated with a signi cantly decreased probability of membership in the non-pathogenic serovar group. This indicates that isolates allocated to the non-pathogenic serovar group may be avirulent based on this vta1 analysis. Similarly, a signi cantly increased probability of harboring vta1 was observed in the highly pathogenic serovars 1 and 5. Based on only the above analysis, the virulences predicted by the serovar and vtaA1 analyses were consistent. However, all 21 highly pathogenic isolates from serovar 10 were vtaA1 negative, which indicates that the highly pathogenic serovar 10 isolates may be avirulent. Thus, the results of virulence prediction by the serovar and vtaA1 analyses were in opposition. The correlation between serovars and VGs varied greatly among different serovars, even if the isolates belonged to the same pathogenic serovar group. For example, serovar 1 was only positively associated with vta1, while serovar 5 was positively associated with 9 VGs. Although the average number of VGs in the three pathogenic serovar groups was similar, the highly pathogenic serovars had a signi cant positive association with 4 VGs, the moderately pathogenic serovars had a signi cant positive association with 2 VGs, and no VGs had a positive association with non-pathogenic serovars. A previous study showed that G. parasuis MLST STs can be classi ed into two clades, with clade one almost completely containing avirulent or attenuated STs, and clade two mainly containing virulent STs (Mullins et al., 2013;Wang et al., 2017). In the current study, the detection frequency of VGs in clade two was much higher than that in clade one. While all clade two isolates were vtaA1 positive, only 30% of clade one isolates were vtaA1 positive. We found a signi cant positive correlation between clade two and 9 VGs. Based on the VG analyses, it appears that isolates belonging to clade two are more virulent than isolates belonging to clade one. Overall, our results show that VG analyses may be a supplementary method for accurately allocating serovars or genotypes of G. parasuis into different pathogenic groups.

Conclusion
We analyzed the serovars and the VGs of 117 G. parasuis isolates from the nasal cavity of live piglets in the south of China. The distribution of serovars and VGs was distinct from previous studies, and we discovered considerable genetic heterogeneity among nasal isolates. Our results complement epidemiological data of G. parasuis and aid our understanding of the extreme genetic diversity associated with this important bacterial species.

Declarations
Authors Contributions All authors contributed to the study conception and design. Ling Peng conceived the experiments and wrote the paper. All authors performed the experiments. All authors read and approved the nal manuscript.
Funding This work was supported by the special projects in key elds of colleges and universities in Guangdong Province and the Department of science and technology of Guangdong Province(grant numbers 2014A020208140) Con icts of interest The authors declare that they have no con ict of interest.
Data availability All data generated and analysed during this study are included in this published article.
Consent to participate Not applicable.
Ethics approval All applicable international, national, and/or institutional guidelines for the care and use of animals were followed.
Consent for publication All authors give consent for publication.
Code availability Not applicable. Figure 1 The Neighbour-joining tree based on the MLST target sequences of 43 G. parasuis isolates