Association between iscR-based phylogeny, serovars and potential virulence markers of Haemophilus parasuis

Isolates and culture conditions

A total of 148 H. parasuis strains were used in this study, of which 145 strains were isolated from clinical samples sent to our diagnostic center from September 2009 to January 2017, and the other three strains were isolated from nasal cavity (Dataset S1). The isolates were originated from 125 independent pig farms in three provinces of eastern China. A sterile loop was used to inoculate TSA plates (supplemented with 2.5% bovine serum and 0.05% NAD) from samples of nasal cavity, lung, spleen, joint, brain, pericardium liquid, pleural effusion, and seroperitoneum. The plates were kept in a 37 °C incubator under humidified 5% CO₂ for 24–72 h until the colony appears. Unequal number of colonies (from four to seven) was picked based on the number of samples, isolation sites or morphology of the colony for each independent farm, and each selected colony was inoculated on another TSA plate for further purification. Then, the isolates were transferred from TSA plates to TSB broth (supplemented with 2.5% bovine serum and 0.05% NAD) for bacteria enrichment.

Serovar determination

Genome DNA of H. parasuis was extracted by digestion with proteinase K, and the isolates were identified by a species specific PCR test (Angen et al., 2007). Serovars of the isolates were determined by the two newly developed molecular serotyping methods, which were faster, more sensitive, and more specific than the traditional IHA (Howell et al., 2015; Jia et al., 2017). Using of the two molecular serotyping methods in combination allows the differentiation of all the 15 serovars. If all the picked colonies from one independent farm were determined to be the same serovar, then one of colonies was selected to be the representative. If more than one serovar were detected in a farm, then one representative for each different serovar was selected for further analysis. The designation of each clinical isolate includes information of province (ZJ for Zhejiang province), year (09 for year of 2009), sequence number (01 for the first isolates of the year) of isolation, and serovar (S1 for serovar 1 etc., Su for unknown serovar) of the isolate.

IscR sequencing and analysis

Primers for amplification of complete open reading frame of iscR gene were designed based on the iscR gene sequence of the strain SH0165 (GenBank accession number: CP001321) (Xu et al., 2011), and listed in Table 1. Amplicon of each isolate was purified and cloned into pMD 19 T vector, and send to TSINGKE Biological Technology Co., Ltd for gene sequencing. IscR gene for each isolate was sequenced twice in both the forward and reverse direction. IscR sequences of 15 reference strains (GenBank accession numbers: KC795297, KC795318, KC795335, KC795364, KC795350, KC795383, KC795404, KC795422, KC795440, KC795459, KC795480, KC795495, KC795512, KC795532, KC795549) of H. parasuis were originated from a previous study (Howell et al., 2013). Alignment of the iscR sequence was prepared with MEGA 7.0, Lasergene 7 or DNAMAN 9. Phylogenetic tree was decorated with an online tool iTOL v4.2.3 (https://itol.embl.de/).

Clinical backgrounds and detection of PVMs

Isolates from systemic sites such as spleen, brain, pericardium liquid, pleural effusion, seroperitoneum, and joint were considered to be systemic ones, and those from lungs and nasal cavities were considered to be respiratory ones and nasal ones, respectively. To further confirm the population structure and its relevance with clinical phenotype, 12 PVMs from five pieces of literatures were selected. Primers of one of the selected PVMs (hsdS) were newly designed in this study (Table 1), and primers of other selected PVMs are the same with that in the literatures (Sack & Baltes, 2009; Wang et al., 2011; Yu et al., 2014; Howell et al., 2017; Galofre-Mila et al., 2017). The presence or absence of the selected PVMs in all the isolates were determined by PCR. Two or more strains from the same farm with any difference in respect of serovar, iscR clade, presence or absence of any PVMs were considered to be different isolates. The differences in average number of PVMs between serovars were determined by One-way ANOVA with SPSS Statistics 17.0. Differences were considered statistically significant when P < 0.05.

Accession numbers

The nucleotide sequences of iscR for the 148 isolates were submitted to GenBank. The accession numbers for the individual CDSs are MH982282–MH982429.

Separation of clinical isolates into two deeply divergent groups by iscR-based phylogenetic analysis

Iron-sulfur cluster regulator gene was presented in all 148 isolates used in this study with the consistent open reading frame size of 447 bp. Phylogenetic sequence analysis of iscR amino acid sequences showed that the isolates were divided into two deeply divergent clades, which were designated as clade I and clade II (Fig. 1). Dominant sequence type (ST1 and ST6) in each clade were further aligned with iscR of other Gram-negative bacteria, and simultaneous amino acid substitutions at highly conserved residue 29, 68, 77 made clade I H. parasuis an evolutionarily distant lineage from this and other Gram-negative species (Fig. 2A). Amino acid sequence comparison revealed nine sequence types from the clinical isolates, and amino acid substitutions at residue 29, 68, 70–71, 77–78, 140 took place at 100% or nearly 100% between iscR of the two clades (Fig. 2B). Four iscR protein sequence types were identified from the 15 reference strains, of which three were identical to those from the clinical isolates in this study (Fig. S1).

Of all the isolates used in this study, clade I isolates were more prevalent (62.2%) than clade II isolates (37.8%). A higher proportion of isolates were systemic ones in clade I (44.6%) than that in clade II (30.4%) (Fig. 3A). No tissue tropism was seen in respect of iscR-based phylogeny, as isolates from all the systemic sites can be found in both clades (Fig. 3B). It is noteworthy that 93.8% of the isolates from pericardial effusion are belong to clade I (Fig. 3B).

Clear correlation between serovar and iscR-based phylogeny

The two molecular serotyping methods allow the assignment of 144 (97.3%) isolates to 13 serovars except serovar 3 and 15, and 4 isolates (2.7%) could not be assigned to any of the 15 serovar known so far. Serovar 4 (18.9%) and serovar 12 (18.9%) were the most prevalent followed by serovar 13 (16.2%) and serovar 5 (13.5%) (Fig. 4A). Isolates within most of the serovars tend to be distributed in the same clade. Serovars 4, 5, and 14 were dominated by clade I isolates, while serovars 1, 2 and 7 were dominated by clade II isolates (Fig. 4A). Although slightly higher proportion of serovar 12 and 13 isolates were found in clade I, the distribution of isolates in the two serovars was relatively equal in both clades. For the remainder of the serovars, the number of isolates was too low to be able to restrict them to one clade.

Systemic isolates were found in 10 of the 13 serovars identified. Of all the serovars with not less than five isolates, systemic isolates were more common in serovar 13 (54.2%) and serovar 12 (50%), and similar proportion of systemic isolates were found in serovar 1, 4, 5 ranging from 35% to 37.5%, which were followed by serovar 7 (20%) and serovar 14 (14.3%) (Fig. 4B). These data indicate that virulence varies among serovars, and serovars 13 and 12 are more strongly linked to systemic infections.

Presentation of PVMs showed obvious correlation with iscR-based phylogeny and serovar

To further confirm the population structure and its relevance with clinical phenotype, distributions of 12 previously identified PVMs were investigated in all the 148 isolates. OmpP2 and vtaA genes could be detected in 100% and 98.6% of the isolates, respectively. Only four isolates were identified to be avirulent by the vtaA method, of which one was isolated from joint of diseased pig. Hence, presence of ompP2 and vtaA gene showed low resolution of the isolates, and the other 10 PVMs were selected in the following studies.

The positive rates of most of the PVMs were similar among serovars (4, 5, 12, 13, and 14) which were dominated by isolates in clade I and among serovars (1, 2, and 7) which were dominated by isolates in clade II. The positive rates were clearly different when compared between serovars of one group (serovars 4, 5, 12, 13, and 14) and the other group (serovars 1, 2, and 7). Most of the PVMs (8/10) were more prevalent in clade I than in clade II (Fig. 5). Of all the serovars with not less than 5 isolates, serovar 5 isolates carried the most PVMs with the average number of 7.35, while serovar 1 isolates carried the least with an average number of 1.5. Serovars dominated by clade I isolates were found with presence of remarkable higher average number of PVMs than those dominated by clade II isolates (P < 0.05) (Fig. 6). Interestingly, average number of PVMs in serovar 12 and 13 were similar to the serovars dominated by clade I isolates (Fig. 6), although the distribution of isolates of the two serovars was relatively equal in both clades. In this study, only one isolate was found with the presence of all the 10 selected PVMs (Fig. 7).

Higher strain diversity in isolates from multi-strain infected farms

A total of 34 PVMs profiles were identified based on the presence or absence of the 10 selected genes (Table 2). The most prevalent profile was shared by 39 isolates, followed by 22 isolates in profile 20. More than half of the profiles (58.8%) were represented by only one isolate, which suggests high gene heterogeneity in accessory genome among isolates. Of the six profiles with no less than five isolates, systemic isolates were most prevalent in profile 27 (4/5) and least prevalent in profile 29 (0/10). Similar proportion of systemic isolates (29–46%) were observed in other profiles (Table 2). Moreover, 10.2% of the systemic isolates do not carry any of the 10 selected PVMs (Table 2), and all of which were distributed in clade II with the exception of one serovar unknown isolate (Fig. 7). Neither presence of any of the selected PVMs nor the number of PVMs carried by the isolates showed correlation with clinical phenotype of systemic infection (Fig. 7).

Of all the 125 independent pig farms, multi-strain (two or three) infection was found in 16% of the pig farms and single strain infection was found in the other 84% pig farms. An average of every 4.4 isolates from the single strain infection farm produce one PVMs profile, while every 2.2 isolates from the multi-strain infection farm produce one PVMs profile (Table 3). Hence, multi-strain infection in one farm resulted in more diverse PVMs profiles in those isolates, which indicates that frequent gene transfer among isolates took place in multi-strain infected farms.

Potential gene exchange at the capsule locus

To study the possible recombination between strains at the capsule locus, all isolates were clustered based on pattern similarity of PVMs. Two distinct clusters were identified, and were designated as cluster A and cluster B (Fig. 8). Most of the isolates (92.1%) of serovar 4, 5, 12, 13, and 14 harbored more than five PVMs, and were distributed in cluster B. Most of the isolates (78.3%) of serovar 1, 2, and 7 only have less than two PVMs, and were listed in cluster A (Fig. 8). Coincidently, clear separation of isolates from those serovars (4, 5, 12, 13, and 14) and other serovars (1, 2, and 7) was observed in iscR-based phylogeny analysis (Fig. 4A). It indicates that these two clusters were possibly originated from two independently evolutionary lineages, which may have branched off from each other for a long evolutionary period.

Recombination event may take place when two or more strains co-exist in one farm, especially when in the same infected organ or commensal site. A small portion of isolates (21.7%) of serovar 1, 2, and 7 such as ZJ1510-S1 and ZJ1310-S7 were divided into cluster B. Similarly, a small portion of strains (7.9%) of serovar 4, 5, 12, 13, 14 were listed in cluster A, for instance, ZJ1604-S4, ZJ1207-S13, ZJ1405-S13, and ZJ1424-S13 etc., (Fig.8). All these isolates harbored PVMs characteristic that do not match the majority of isolates of the same serovar, but consistent with PVMs characteristic of serovars which were mainly distributed in the other cluster. It indicates that these isolates may have undergone gene exchange between one serovar group (1, 2 or 7) and the other serovar group (4, 5, 12, 13, and 14) at or around the site of the capsule locus. Hence, isolates in the same serovar can vary greatly in terms of PVMs, which could lead to differences in the pathogenicity and immunogenicity within one serovar. It would be more difficult to identify in this study when exchange of capsule locus take place between serovars with similar PVMs characteristics.