Molecular Epidemiology Characteristics of Brucella Abortus Strains in China and Their Relationships With World Lineages

Background Brucella abortus is a facultative intracellular Gram-negative bacterium that causes chronic persistent infections in humans and livestock. In this study, conventional bio-typing, multiple-locus variable-number tandem repeat analysis (MLVA), and whole-genome sequencing-single-nucleotide polymorphism (WGS-SNP) were used to investigate the molecular epidemiology characteristics of Brucella abortus strains in China and their relationships to world lineages. Results A total of 100 strains were collected from 1953 to 2013, suggesting that B. abortus circulated in China in the past ve decades. Moreover, most strains were mainly distributed in the Northwest areas, suggest that provinces in the Northwest were a dominant epidemic area of this disease. During this period, seven biovars were found, indicating that B. abortus had a high diversity of biovars and it is also a potential reason for the disease ongoing spread in the Northern provinces. Strains have high genetic diversity, and bruce07 is the most helpful locus for genotyping of this population. Moreover, 17 MLVA-11 genotypes were found; 13 of them are of known genotypes and four are unassigned genotypes, indicating that B. abortus in this study had several geographic origins. Still, strains from unassigned genotypes may originate from China. Many shared MLVA-16 genotypes were observed in strains from the same provinces in Northern China, which conrmed a B. abortus brucellosis outbreak within Northern regions. WGS-SNP analysis showed that eight Chinese strains formed a ladder-like phylogram (C. (cid:0) ) with strains from nine countries, including Uganda, Iraq, Russia, Georgia, Spain, Italy, Egypt, Mongolia, and China; suggest that strains were introduced to these countries from a single source. Conclusions Chinese B. abortus strains had high biovars and genetic diversity as well as represent characteristics of multiple geographic origins, and B. abortus strains from several mainly epidemic areas were closely related to strains from Russia and Mongolia; frequent animal (cattle) trade and exchanges may promote this process. We will provide new and valuable information to strengthening surveillance and control of B. abortus brucellosis in China.

back, and/or loss of appetite, as well as resulting in signi cant public health risk [13]. In the infected animals, abortion, retained placenta, infertility, orchitis, epididymitis is common symptoms and due to reduces livestock production and reproduction cause hung economic loss [14].
Our previous study showed that B. melitensis is a latent "travel bacterium" that spread and expanded from North China to South China [15]. However, study on the molecular epidemiology characteristics of B. abortus countrywide is rare [16]. In Northern China, where the traditional agropastoral areas with more developed animal breeding industries are located, the B. abortus brucellosis seroprevalence was greater than 10%; the seroprevalence of brucellosis in Southern China reached only 5.5% [17]. Despite the relatively low seroprevalence (2.8%) in a serological survey of yaks, the estimated economic losses were substantial in Tibet. With no control program for brucellosis in yaks, the total annual economic loss in the study area was estimated at US$521, 043, mainly resulting from abortions and reduction of the milk and meat yield [18]. Therefore, the molecular epidemiology of brucellosis caused by B. abortus at the national level is an urge.
Although bio-typing was useful for Brucella strain identi cation and species differentiation, it is timeconsuming and di cult to interpret due to a lack of standardization of the typing reagents [19]. Moreover, multi-locus variable-number tandem repeat assays (MLVA) can providing a ne-scale resolution among isolates and allowing the determination of the source of infection, geographical origin, and spread pattern of disease [20,21]. The whole-genome sequencing-single-nucleotide polymorphism (WGS-SNP) is used to establish phylogenetic relationships of Brucella isolates [22,23]. In this study, 100 strains were obtained from samples collected between 1953 and 2013 from various host animals throughout the countrywide in China. Our study was aimed to determine the distribution regions, genotyping characteristics, geographic origins, and phylogenetic relationships of B. abortus strains at the global level.

Bacterial source and identi ed
The MLVA-16 characteristics, panel1, MLVA-11 genotypes, species/biovars, hosts, location, and year of B. abortus strains from China were collected, and Microsoft Excel (Microsoft, Redmond, CA, U.S.) was used for data cleaning. A total of 100 strains were collected from 14 different provinces from animals and humans, which of 65 were found in cattle, 20 in humans, 7 in sheep, 4 in yak, 3 in goat, and for 1 strain the host is unknown. Moreover, 36 strains previously published from Inner Mongolia [24] were involved in this study. Species/biovars of all 100 strains were identi ed based on standard bio-typing procedures [25], the speci c identi cation methods were identical to previously published [26,27]. Furthermore, AMOS-PCR (B. abortus bv. 1, 2, and 4, B. melitensis, B. ovis, and B. suis bv. 1) [28] was used to verify further B. abortus bv. 1, 2, and 4.

Analysis of the MLVA data
Hunter-Gaston diversity index (HGDI) [33] was applied to evaluate the genetic diversity of B. abortus strains as well as the resolution of each locus in the MLVA approach in this study. MLVA-16 data (Table   S1) in all strains were analyzed by the BioNumerics version 5.1 software (Applied math, Belgium) based on the categorical coe cient and unweighted pair group methods. MLVA-11 was used to investigate the geographical origins between our isolates [34,35] and 1,482 isolates from the MLVA bank (http://microbesgenotyping.i2bc.paris-saclay.fr/databases) (Table S2). Subsequently, the minimum spanning tree was constructed by the goeBURST algorithm using the Phyloviz software v 2.0 [36].
Investigation on genetically related among strains from our reported (n = 100) (Table S1) and other regions of worldwide (n = 1,525) based on the MLVA-16 data was performed (Table S3). Moreover, phylogenetic analyses of 62 B. abortus (Table S4) genomes retrieved from GenBank were performed, the phylogenetic tree was constructed based on the PHYML (PHYlogenetic inferences using Maximum Likelihood) software with parameter default values [37].

Worldwide Phylogenetic Analysis Based On Both Mlva-16 And Wgs-snp
In present study, 1,625 strains were sorted into three groups (A-C), and Chinese strains fell into group A and clustered together with strains from Italy, Kazakhstan, and Brazil, but had no sharing of the MLVA-16 genotype was observed (Fig. 3). Whole-genome SNP analysis showed that 65 B. abortus strains were grouped into seven clades ( -) (Fig. 4) ladder-like phylogram. However, most strains (n = 8) from China were found in clade and were the most similar to strains from Russia and Mongolia (Fig. 4).

Discussion
B. abortus infection was highly endemic in dairy herds in China [17]. The comprehensive molecular epidemiological and phylogenetic analysis of the B. abortus strains nationwide are useful for improving the surveillance and control program to determine the risk factors associated with brucellosis in humans.
In this study, 100 B. abortus were collected from 1953 to 2013, indicating that B. abortus strains have circulated in China for the past several decades, but the strain numbers were lower than those of B. melitensis [38]. This observation is in agreement with a report that B. abortus was a secondary pathogen agent for brucellosis in China [9]. The seven biovars were observed in this study, and B. abortus bv. 3 was dominant species in China, suggesting that B. abortus in China has had a high diversity of species/biovars, which poses a potential challenge for animal vaccination. The 85% strains were distributed in traditional and domestic animals breeding developed provinces, providing helpful information for disease control and prevention. However, the distribution range of this species is different from that of the B. melitensis strains, which cover all mainland regions in China [39]. Similarly, the most frequently recorded B. abortus biovar is biovar 3 in the Republic of Kazakhstan. More than 90% of the overall B. abortus samples were isolated from the northern regions of East and West Kazakhstan [40]. Five kinds of hosts were observed in this population, and 65% of strains were from cattle. This indicated a narrower host spectrum than that of B. melitensis [41], which is a signi cant impact factor for strains with similar geographic distribution and is closely related to the low virulence of B. abortus strains. In Italy, the strains identi ed from cattle showed a high prevalence of B. abortus bv. 3 isolates (84.5%), followed by B. melitensis bv. 3 (9.9%) and B. abortus biovars 1 and 6 (5.5% and 0.1%, respectively) [42]. In Africa, B. abortus bv. 3 is the most commonly isolated strains in cattle [ [43]. Moreover, there were no strains from swine and wild animals; further pathogen surveillance in other hosts is a priority.  [48], suggesting that the strains described in this study were introduced from multiple geographic regions.
A total of 17 MLVA-11 genotypes were observed; the higher diversity of MLVA-11 genotype was observed in strains from Inner Mongolia, Xinjiang Province, Hebei Province, and Sichuan Province, suggesting that animal trade often occurred among these regions in the past. The seven shared MLVA-11 genotypes (72, 75, 82, 83, 181, 210, and 328) comprising strains from 2 to 10 countries suggest that the strains from each shared genotype had a common source, indicating that the B. abortus from this study had multiple geographic, cross-border transmission, probably due to exchange and trade of infected animal. Trade ties and livestock exchange between countries have a long history; archeological ndings unraveled trade relations between the nomadic people who inhabited Central Asia and China long before the Common Era (C.E.) [49]. MLVA-11 genotype 72 was shared by strains from six countries, including France, Germany, Italy, Kazakhstan, Portugal, and China. MLVA-16 comparison analysis showed that Chinese strains were clustered together with Italy, Kazakhstan, and Brazil. A previously reported that, 4,500 years ago, seasonal nomadic pastoralist routes were formed from modern Southern Kazakhstan to Xinjiang Uygur Autonomous Region, covering more than 70% of the high-mountain route of the silk road [50]. However, there is no trade-in live cattle between Kazakhstan and China in modern history, but actively developing livestock products [51]. Furthermore, four strains from Sichuan Province had three MLVA-11 genotypes (83, 77, and 82), two of which were shared (82 and 83) with strains from other regions, and one (77) is exclusive of strains from Sichuan Province. Although it was a historical area of B. abortus brucellosis, it had no shared genotype 72 [9]. The investigation of serology and bacteriology of B. abortus infection in this region is a priority. However, 10 single genotypes were exclusively found in Chinese strains, indicating that these strains originated from China lineages, but further investigation on the origin of the strains is recommended.
In this study, the 19 shared MLVA-16 genotypes included 54 strains, with each genotype representing 2-8 strains, and 18 out of 19 shared genotypes consisted of strains of the same province. Only one shared genotype (GT48) was present in ve strains, and these were obtained from four different provinces, suggesting that the majority of brucellosis cases were outbreak epidemic within respective regions. This conclusion is in agreement with the geographic distribution of the strains and outbreak cases mainly focus on Northern China. Moreover, the remaining 46 independent genotypes of each single strain indicated that 46% (46/100) cases had epidemiology unrelated or sporadic characteristics [31]. Strengthening the surveillance of B. abortus animal infection applied serology, bacteriology, molecular assays in northern regions, China helps better understand the epidemiology of brucellosis.
The sixteen Chinese strains were divided into ve clades by whole-genome SNP analysis, con rming that strains from this study had multiple ancestor strains. Remarkably, clade formed a ladder-like phylogram and consisted of strains from nine countries, including Uganda, Iraq, Russia, Georgia, Spain, Italy, Egypt, Mongolia, and China from a single ancestor. The B. abortus isolates from Italy are substantially different from those found in Europe and North America and are more closely related to strains from the Middle East and Asia [52]. Furthermore, most Chinese strains (n = 8) were the most similar to Russia and Mongolia; they were geographically close. They revealed that strains from a common ancestor were continuously circulating among these countries. Human brucellosis seroprevalence among rural people in Mongolia is high [53], and Mongolian B. melitensis isolates had high genetic similarity with Chinese strains, likely due to geographical proximity [54]. Moreover, some strains from Mongolia had closely related MLVA genotypes to strains from Russia [55]. In the North Caucasian Federal District, the largest number of new human brucellosis cases was noti ed in the Republic of Dagestan (59.3%) and the Stavropol Territory (27.4%), and the true prevalence of brucellosis greatly exceeds the o cial statistical data [56]. The frequent spread of these lineages from one country to another due to long-term trading partnerships between the three countries is a likely explanation for the data. Otherwise, due to the uncontrolled introduction of the agents via humans, infected animals, semen, and vectors have a high risk of B. abortus infection [57]. In endemic countries, combining the serodiagnosis and bacteriology, and molecular diagnosis for surveillance of brucellosis is essential. Molecular genotyping should be systematically applied to support control plans for control of brucellosis in China.
Moreover, our study has some limitations. First, the strains were collected from previous studies that might have been in uenced by many aspects, such as the local economic situation and the diagnosis and surveillance status of brucellosis. Second, due to variability in the number of strains collected among different provinces and for different years in this study, a genome analysis of strains from more regions and hosts is recommended. Third, no related epidemiology data were collected, and analysis of animals moves, and exchange is lacking. Therefore, a whole-province survey on human and animal infections with B. abortus should be initiated.

Conclusion
In this study, a nationwide molecular analysis of B. abortus strains were performed. Our research showed that there was a high diversity of biovar and MLVA-11 genotypes among this population. B. abortus strains exhibited multiple geographic origin characteristics and the predominant genotype was shared with strains from many countries. Some strains from main epidemic areas were similar to strains from Russia and Mongolia. These have been traditionally countries of cattle and sheep breeding industries.
Our work contributes to a better understanding of the epidemiology of B. abortus brucellosis in China and also provides invaluable information that could be helpful to devise control strategies for the disease.

Consent to publish
Not applicable Availability of data and materials All data generated or analyzed during this study are included in this published article and its supplementary information les.

Competing interests
The authors have no con icts of interest.

Funding
This study was supported by the National Key R&D Program of China, grant number 2018ZX10734401, 2018ZX10734404, and 2017ZX10303401. The funding agencies had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Authors' Contributions
LZG and ZZZ performed strains collected and analysis, WXM and LZG performed genotyping and cluster analysis and drafted the manuscript; WM conducted epidemiological investigations and data analysis; CBY and LZG participated in the design of the study and critically reviewed the manuscript; WXM and LZJ participated in the design of the study and managed the project. All authors read and approved the nal version of the manuscript.