Epidemiology of PDCoVs
PDCoV was first identified by Deltacoronavirus specific-PCR in rectal swabs of pigs (10.1 %, 17/169) with unknown healthy status in Hong Kong . Then, PDCoV emerged in United States, China, South Korea and Thailand [5, 17, 18]. In most of studies, excluding PEDV, TGEV and porcine rotavirus, PDCoV, as an important enteric pathogen, was detected in clinical samples from pigs with diarrhea [9–12, 23]. In addition, it was confirmed experimentally that less than two-week old piglets were susceptible to PDCoV, which caused mild to moderate diarrhea as well as macroscopic and microscopic lesions in small intestines of conventional piglets (5-day-old), and severe diarrhea, vomiting, fecal shedding of virus, and severe atrophic enteritis in gnotobiotic pigs (11- to 14-day-old) [2, 3]. The data further confirmed that PDCoV were enteropathogenic in pigs. Meanwhile, PEDV or rotavirus showed higher detection rates in PDCoV-positive samples compared with other TGEV and rotavirus [8, 13–15, 24]. As shown above, co-infection of PDCoV and PEDV occurred in nursing piglets (Table 1), indicating that the diarrhea-related pathogens were quite complex clinically and not easy to control in the field. Moreover, in the two recent studies, PDCoV was shown to have higher infectivity in sows with diarrhea (81.0 %, 34/42) than nonclinical counterparts (23.5 %, 4/17) [8, 15], which might imply that sows often carry PDCoV. And further, it could result in the transmission of PDCoV from sows to the foetus and even newborn piglets, although the pathogenesis mechanism of PDCoV remains unclear.
To further understand the origin of PDCoV, some retrospective studies were made using PCR and enzyme-linked immunosorbent assay (ELISA) [24–26]. In Dong et al.  study, 2 of 6 samples collected from Anhui Province of China in 2004 were positive for PDCoV, up to now, it was the most ancient time for the detection of PDCoV in China. Meanwhile, PDCoV could date back to August 2013 in United States, where only 3 PDCoV-samples were detected using PCR in archived samples . As for serology of PDCoV, anti-PDCoV IgG antibodies could date back to 2010 using an indirect anti-PDCoV IgG ELISA based on the putative S1 portion of the spike protein . The above studies indicated that PDCoV could have circulated in China at least since 2004 and in United States since 2010. Maybe, due to limted samples in the present study, we did not detect PDCoV in pig samples collected from Guangdong province between 2012 and 2014. Although Asian leopard cat coronavirus (GenBank accession no. EF584908) was close to PDCoV in the phylogenetic trees (Figs. 1 and 2), in the future, more epidemiological surveys should be warranted to uncover the origin of PDCoV.
At the territory of China, Southern China mainly includes Guangdong province, Hainan province and the Guangxi autonomous region. Although molecular detection of PDCoV was performed in these regions [23, 24], little information was available on PDCoV prevalence. In a study by Chen et al. , an overall positive-PDCoV rate of 23.4 % (15/64) was obtained in all samples collected from Guangdong, Shanxi and Hubei provinces. However, more detailed data of PDCoV was not available in Guangdong province. Meanwhile, in the study from Dong et al. , only four archived samples from the Guangxi autonomous region were examined, but all negative for PDCoV. In this study, we demonstrated that PDCoV circulated and was co-infected by PEDV on those swine farms in Guangdong province, Hainan province and the Guangxi autonomous region, which further contributed to the epidemiology of PDCoV in these regions despite the relatively low prevalence.
Genetic diversity of PDCoVs
The first two reported full-length PDCoV genome sequences (HKU15-44 and HKU15-155) were 25, 437 nt and 25, 432 nt in length, respectively , and they had 99.1 % nucleotide similarity with each other. Moreover, further sequence alignment showed a 3-nt (TAA) insertion in the S gene and a 3-nt (TTA) insertion in the 3’ untranslated region (UTR) of HKU15-44 [1, 5]. During the past 3 years, PDCoV-associated swine enteric disease was paid great attention in the major pig producing countries, especilly United States and China. Up to May 2016, more than 30 complete PDCoV genome sequences were published in the GenBank database. All were generated in China and United States except for one sequence from South Korea and three sequences from Thailand [8–19]. The Korean strain, KNU14-04, had 25, 422 nt in length, with similar genome features (a 3-nt insertion in the S gene with 3, 483 nt and a 3-nt insertion in the 3’ UTR, respectively) to all American strains and the Chinese strain (HKU15-44) . Comparing complete genomes of the remaining Chinese strains to the American, Thai and Korean counterparts, CHN-HB-2014, CHN-JS-2014, PDCoV/CHJXNI2/2015, CH/Sichuan/S27/2012, CH/SXD1/2015 and P23_15_TT_1115 only had the 3-nt (AAT) deletion in the S gene (3, 480 nt) [14, 23, 24], while CHN-AH-2004 only had the 3-nt (TAA) deletion in the 3’ UTR [5, 24]. In the present study, 3-nt insertion was not found in UTR for our five obtained PDCoV (data not shown). Moreover, two additional unique features including a 6-nt (TTTGAA) deletion in the nonstructural protein (nsp) 2 region and a 9-nt (GCCGGTTGG) deletion in the nsp 3 region were also found in CH/Sichuan/S27/2012 . However, for Thai viral isolates, they owned one additional unique nucleotide (C) insertion in the 3’ UTR [17, 18]. The biological significance of these naturally occurring deletions or insertions in PDCoV biology and pathogenesis warrants further investigations.
In this study, five S and five N gene sequences, respectively, were obtained to evaluate wherever genetic diversity of PDCoVs existed in southern China. Our results showed that these five S and five N gene sequences were more closely related to Chinese strains, and all clustered together in the phylogenetic tree (Table 2, Figs. 1 and 2). However, CH/GD01/2015 and CH/GD02/2015, reported in this study, originated from the same pig farm in Guangdong province, but had 48 nt and 12 nt differences in the S and N genes, respectively. The observed 48 nucleotide changes in the S gene made these viruses differ by 25 amino acid residues (Additional file 2: Table S2). For the N gene, the 12 nucleotide changes among these viruses resulted in 3 amino acid substitutions (Additional file 2: Table S2). Among them, 18 of 25 amino acid differences occurred at the first two-third parts of S gene. Interestingly, in spite of amino acid mutation, both S and N protein retained almost consistent amino acid properties (especially pH value) (Additional file 2: Table S2). Future study will address important roles of these polymorphisms in viral replication and pathogenesis. In addition, they were divided into two distinct small branches (Figs. 1 and 2). These findings suggested that PDCoVs in southern China have diverged from a common ancestor. Despite the emerging genetic diversity, overall, PDCoV prevalence is still largely restricted by the territory as demonstrated in Figs. 1 and 2.
For the two enteric coronaviruses (PEDV and TGEV) in pigs, the recombination events were often detected. However, most of them were from intra-recombination [27–30]. Recently, only one emerging recombinant/chimeric virus (named swine enteric coronavirus, SeCoV) was discovered in swine feces and resulted from inter-recombination of PEDV and TGEV, which had a TGEV backbone and a PEDV spike gene [31, 32]. In this study, there were no any possible recombinant events occurring in PDCoV strains. Maybe, the number and length of our obtained PDCoV sequences were limited. In the following study, the recombination event of PDCoV warrants further attentions.