Virulence, Antibiotic Resistance, and Phylogenetic Relationships of Aeromonas spp. Carried by Migratory Birds in China

This study aimed to evaluate antimicrobial resistance, virulence, and the genetic diversity of Aeromonas isolated from migratory birds from Guangxi Province, Guangdong Province, Ningxia Hui Autonomous Region, Jiangxi Province, and Inner Mongolia in China. A total of 810 samples were collected, including fresh feces, cloacal swabs, and throat swabs. The collected samples were processed and subjected to bacteriological examination. The resistance to 21 antibiotics was evaluated. A phylogenetic tree was constructed using concatenated gltA-groL-gyrB-metG-PPSA-recA sequences. Eight putative virulence factors were identified by PCR and sequencing, and a biofilm formation assay was performed using a modified microtiter plate method. In total, 176 Aeromonas isolates were isolated including A. sobria, A. hydrophila, A. veronii, and A. caviae. All isolates showed variable resistance against all 16 tested antibiotic discs, and only one antibiotic had no reference standard. Six kinds of virulence gene markers were discovered, and the detection rates were 46.0% (hlyA), 76.1% (aerA), 52.3% (alt), 4.5% (ast), 54.0% (fla), and 64.2% (lip). These strains were able to form biofilms with distinct magnitudes; 102 were weakly adherent, 14 were moderately adherent, 60 were non-adherent, and none were strongly adherent. Our results suggest that migratory birds carry highly virulent and multidrug-resistant Aeromonas and spread them around the world through migration, which is a potential threat to public health.


Introduction
Wild animals are one of the main reservoir hosts of bacterial antibiotic resistance genes and play a key role in their transmission [1]. Among wild animals, those of the class Aves occupy a wide range of niches all over the world. Birds, especially waterfowl, have long-distance migration, complex physiological characteristics, unique diets, and lifestyles, and live in diverse natural environments such as remote mountains and lakes [1]. As in all vertebrates, trillions of microorganisms live in birds' intestines [2]. The most frequently studied bacteria are Escherichia coli, Salmonella enterica, and Campylobacter jejuni [3], and little attention has been paid to Aeromonas spp. The harm inflicted by Aeromonas on migratory birds cannot be ignored [4]; consequently, a comprehensive study on Aeromonas in migratory wild birds is urgently needed. subcultured on BHI agar plates and incubated for 16-18 h at 36 ± 1 • C [15]. The NMIC/ID four panel of the BD Phoenix TM-100 automated microbial identification system (Becton Dickinson Company, East Rutherford, NJ, USA) was used to determine the sensitivity of Aeromonas to 21 types of antibiotics, shown in Supplementary Table S1. The instrument automatically interprets the results, and the determination of the drug sensitivity results is based on the standards of the American Association for Clinical and Laboratory Standards [23,24]. A single colony of Aeromonas isolates was cultured in 1 mL BHI broth overnight at 36 ± 1 • C, 600 µL enrichment solution and 400 µL 50% glycerol normal saline were added and the sample and stored at -80 • C.

Detection of Virulence Determinants by PCR Assay
The suspension of the Aeromonas isolates was centrifuged for 2 min at 12,000× g; the supernatant was discarded and resuspended in 200 µL sterile deionized water, and incubated in boiling water for 10 min. The tube was again vortexed and centrifuged for 2 min at 12,000× g [8]. The supernatant was transferred to a fresh tube and stored at -20 • C and acted as a DNA template for the subsequent detection of virulence genes. Eight virulence genes were detected by PCR. The primers were synthesized by Kumei Bio Inc and their sequences are shown in Table 1. Each PCR amplification of virulence genes was performed in a reaction volume of 25 mL, containing 12.5 µL Taq PCR MasterMix (2×), 0.5 µL 10 µM primer, 1 µL DNA template, and 10.5 µL ddH 2 O. The template was amplified under the following cycling conditions: pre-denaturation at 95 • C for 5 min, 30 cycles of denaturation at 95 • C for 30 s, annealing at 55-66 • C for 30 s, and extension at 72 • C for 1 min, followed by a final cycle at 72 • C for 5 min. A. hydrophila ATCC 7966 was included as a positive control and ultrapure deionized water as a negative control in the PCR reactions. The PCR-positive products were further confirmed by sequencing.

Detection of Biofilm Formation
Biofilm formation by the Aeromonas isolates was assessed using the microtiter plate method described by Isoken H. Igbinosa, with modifications [25]. Ninety-six-well cell culture plates (Costar, Washington, DC, USA) were filled with 180 µL of BHI and inoculated with 20 µL of Aeromonas isolates grown overnight and standardized to 0.5 McFarland. Plates were incubated for 12 h at 37 • C. Positive control wells contained A. hydrophila ATCC 7699 and the negative control wells contained uninoculated BHI. Studies were done in triplicate for each well. The contents of each well were discarded and the wells washed three times with sterile phosphate-buffered saline (PBS). After airdrying, the wells were stained with 200 mL of 1% crystal violet for 30 min. The wells were carefully washed with PBS to remove the excess stain. Plates were allowed to dry at room temperature. Dye bound to adherent cells was resolubilized with 200 mL of absolute ethanol. The optical density (OD) value of the biofilm was determined at 450 nm using an Enzyme standard instrument (BioTek Instruments Inc, Winooski, VT, USA) and the average OD of each duplicate result was taken, including positive and negative controls. Biofilm formation was confirmed according to the criteria of Afreenish Hassan et al. [26], as shown in Table 2. Table 2. Interpretation of biofilm production.

Phylogenetic Analysis
The primer sequences of gyrB, groL, gltA, metG, ppsA, and recA were synthesized [12] and the PCR products were forward sequenced. The consensus sequence for each gene fragment was determined by the ClustalW2 alignment. Multiple alignments containing the concatenated sequences were straightforward and were performed according to the genomic gene order, gyrB, groL, gltA, metG, ppsA, and recA. All analyzed MLST sequences had the same length (2893 nucleotides) [12]. For phylogenetic analysis, concatenated sequences were aligned and analyzed using MEGA X, the phylogenetic tree was constructed using the neighbor-joining method, and iTOL was used for the phylogenetic tree display and annotation. Bootstrap analyses were performed using 1000 replications for NJ.

Statistical Analysis
The dates were analyzed using SPSS version 25.0. A nonparametric one-way analysis of variance (ANOVA) was used to analyze the results. A p value of < 0.05 was considered statistically significant, while a p value of < 0.01 was considered highly significant.

Isolation of Aeromonas spp.
It was assumed that large, flat, and green single colonies with dark green centers (2-3 mm in diameter) represented Aeromonas. A total of 176 strains of Aeromonas were isolated from 810 samples (feces, cloacal swabs, and throat swabs) from migratory birds from five provinces in China with the isolation rate for each region shown in Figure 1 (the map in Figure 1 was obtained from the USGS National Map Viewer). The results of species identification using the BD Phoenix TM-100 automated microbial identification system demonstrated that the most prevalent species in the fecal samples from migratory birds were A. sobria 109 (61.9%), A. hydrophila 53 (30.1%), A. veronii 9 (5.1%), and A. caviae 5 (2.8%). As shown in Table 3, the main epidemic strains are strains of A. sobria in various regions. The isolation rates of feces, throat swabs, and anal swabs were 38.4%, 24.7%, and 3.2%, respectively. The isolation rates of different types of samples showed a significant difference (p < 0.005).
identification using the BD Phoenix TM-100 automated microbial identification system demonstrated that the most prevalent species in the fecal samples from migratory birds were A. sobria 109 (61.9%), A. hydrophila 53 (30.1%), A. veronii 9 (5.1%), and A. caviae 5 (2.8%). As shown in Table 3, the main epidemic strains are strains of A. sobria in various regions. The isolation rates of feces, throat swabs, and anal swabs were 38.4%, 24.7%, and 3.2%, respectively. The isolation rates of different types of samples showed a significant difference (p < 0.005).      (7) a The relationship between sampling sites and human habitation. A: Distant; B: Adjacent; C: Within.

Resistance Phenotypes of the Aeromonas Strains
As shown in Supplementary Table S2, the drug resistance spectrum of the Aeromonas strains to 16 antibiotics was confirmed; the drug resistance rates for ampicillin and ampicillin/sulbactam were the highest, at 97.7% (172) and 89.8% (158), respectively. The drug resistance rate to cefazolin was 39.2% (69), and those for gentamicin, cefotaxime, piperacillin, colistin, trimethoprim-sulfamethoxazole, chloramphenicol, and tetracycline ranged from 8.0% to 14.8%, while the drug resistance rates for ceftazidime, cefepime, aztreonam, amoxicillin-clavulanate, and ciprofloxacin ranged from 0.6% to 4.0%. The distributions of antimicrobial susceptibility, to the antibiotics tested, of A. sobria, A. hydrophila, A. veronii, and A. caviae isolated from migratory bird samples are presented in Figure 2. Each type of Aeromonas has significant differences in antibiotic response (p < 0.05). A. hydrophila and A. sobria had the most kinds of antibiotic resistance and the highest drug resistance rate, so that A. hydrophila and A. sobria had the most serious drug resistance problem. However, the antibiotic resistance of A. caviae should not be overlooked.

Resistance Phenotypes of the Aeromonas Strains
As shown in Supplementary Table S2, the drug resistance spectrum of the Aeromonas strains to 16 antibiotics was confirmed; the drug resistance rates for ampicillin and ampicillin/sulbactam were the highest, at 97.7% (172) and 89.8% (158), respectively. The drug resistance rate to cefazolin was 39.2% (69), and those for gentamicin, cefotaxime, piperacillin, colistin, trimethoprim-sulfamethoxazole, chloramphenicol, and tetracycline ranged from 8.0% to 14.8%, while the drug resistance rates for ceftazidime, cefepime, aztreonam, amoxicillin-clavulanate, and ciprofloxacin ranged from 0.6% to 4.0%. The distributions of antimicrobial susceptibility, to the antibiotics tested, of A. sobria, A. hydrophila, A. veronii, and A. caviae isolated from migratory bird samples are presented in Figure 2. Each type of Aeromonas has significant differences in antibiotic response (p < 0.05). A. hydrophila and A. sobria had the most kinds of antibiotic resistance and the highest drug resistance rate, so that A. hydrophila and A. sobria had the most serious drug resistance problem. However, the antibiotic resistance of A. caviae should not be overlooked. Olumide Odeyemi [27] determined the statistics of the drug resistance spectrum of Aeromonas spp., and it was observed that isolates were susceptible to amikacin, imipenem, Figure 2. The distributions of antimicrobial susceptibility, to the antibiotics tested, of A. sobria, A. hydrophila, A. veronii, and A. caviae isolated from migratory bird samples are presented. Univariate ANOVA revealed that the difference between groups was highly significant for each antibiotic. Other pairwise comparisons within each group were significant. Olumide Odeyemi [27] determined the statistics of the drug resistance spectrum of Aeromonas spp., and it was observed that isolates were susceptible to amikacin, imipenem, meropenem, and levofloxacin. In this study, the 176 strains showed 31 kinds of drug resistance spectrum. Resistance to more than three types of antibiotic was defined as multi-antibiotic resistance (MAR ≥ 0.25) [28]; as shown in Supplementary Table S3, the multidrug resistance rate was 0.23% (41). Among these samples, there were 28 kinds of drug resistance phenotypes in the north, of which A (48.2%) and B (13.4%) were the main phenotypes. The most resistant strains were resistant to eight types of antibiotics. There were seven kinds of drug resistance phenotypes in the south, with A (41.7%) and B (16.7%) as the main phenotypes. The multidrug resistance rate was 25% (3); these strains were resistant to six kinds of antibiotics at most.

Virulence Determinants of Aeromonas spp.
Six kinds of virulence gene markers were discovered in the 176 isolates, and the detection rates were 46.0% (hlyA), 76.1% (aerA), 52.3% (alt), 4.5% (ast), 54.0% (fla), and 64.2% (lip), as shown in Supplementary Table S4. The ascF-G and act genes were not detected in any isolate. A comparison of virulence gene rates in different regions of China showed that the relative abundance of virulence genes was higher mainly in Dali Lake, Ningxia, and Ordos. The detection rates were higher in A.

Biofilm Formation by Aeromonas spp.
In this study, Aeromonas isolates were categorized into four groups according to the strength of biofilm formation. The weak producers of biofilm, which make up 58.0% (102) of the isolates, were most common among the groups classified; 8.0% (14) demonstrated moderate capabilities for biofilm production, whereas 34.1% (60) did not form significant biofilm. No strains formed strong biofilms, as shown in Table 4. Among the four species of Aeromonas, only 14 A. sobria were classified as moderately biofilm-forming; no A. caviae strains tested formed biofilm. Table 4. Biofilm production of Aeromonas isolates in microtiter plates.

Phylogenetic Relationships
The portions of the six housekeeping genes were successfully amplified and sequenced in all 176 strains, except that the gyrB and recA were not amplified successfully in the Aeromonas strains NM31 (A. sobria) and NM25Z (A. sobria). The phylogeny of the 174 Aeromonas strains was analyzed by constructing a neighbor-joining tree from the 2893-bp concatenated sequences (Figure 3). The tree revealed three major phylogroups, of which A. sobria, A. hydrophila, and A. caviae are the main strains. Among the four Aeromonas species, A. sobria and A. veronii clearly showed a high degree of relatedness, and Aeromonas isolates from different regions also had close genetic relationships. The phylogenetic tree revealed strong nodal support for three major lineages. All the isolates in different branches were easily distinguishable.   In the phylogenetic trees constructed from individual gene trees, the PPSA gene tree has the highest consistency, the gyrB gene tree, groL gene tree and gltA gene tree are generally consistent, while there was a large difference in the recA gene tree and metG gene tree when compared with the concatenated sequence tree. Most remarkably, the phylogenetic tree constructed for the housekeeping gene seems to be related to that of the virulence genes.

Discussion
From a public health perspective, the long-distance migration of waterfowl provides a mechanism for the global transmission of bacterial pathogens [29]. The East Asian-Australasian Flyway (EAAF) is one of four globally recognized flyways for migratory waterbirds, and the geographical location of China means that it plays an important role in the migration route of the EAAF migratory birds [30]. In this study, we focused on the differences among Aeromonas species of migratory birds in different regions of China. One of the main results of the study was that Aeromonas spp. carried by migratory birds are mainly A. sobria, A. hydrophila, A. veronii, and A. caviae, in agreement with Cardoso et al. [4]. In addition to A. sobria, all Aeromonas species identified herein have been related to cases of gastroenteritis in humans, which indicates clinical importance [4]. At the same time, the infection and pathogenicity of A. sobria cannot be overlooked, there are increasing studies that have reported clinical cases of A. sobria infection [31]. This shows that humans pose a serious threat to migratory birds and pose a serious threat to public health.
The second main result of the study is that the isolation rate of Aeromonas spp. in the north was significantly greater than that in the south. Microbial communities are highly dynamic, affected by both internal (e.g., physiological state, sex, breeding status, genetic predisposition) and external (e.g., season, location, diet, social interactions) factors [32]. There are differences in environmental temperature, air humidity, day length, geography, and environment between the south and the north of China [33]. A possible reason for the discrepancy is that the timing of large-scale dissemination of Aeromonas spp. was inconsistent with the time of sampling. Although the climate in the south is more suitable for the growth of Aeromonas, the isolation rate of Aeromonas in the north was higher in this study. The reason for this result is that the migration of migratory birds to the south occurs in January, when the metabolism and reproductive capacity of Aeromonas are low, while the sampling time in the north was in April, when the metabolism and reproductive capacity of Aeromonas are increased. Although the sampling time was different in the north and south, the main epidemic strain was A. sobria, which is inconsistent with the results of Yongxiao Zhu [33].
The third main result of the study is that the isolation rate from cloacal swabs was lower than that of oropharyngeal swabs and the isolation rate from swimming birds was higher than that from wading birds. This may be because of the ubiquitous presence of Aeromonas spp. in surface waters, and the feeding mode of swimming birds. There are many factors affecting the composition and diversity of intestinal bacterial communities of migratory birds, including diet, environment, and season [34].
In this study, the Aeromonas spp. isolated from wild birds in most areas had some drug resistance. Resistance rates differed slightly between regions: the drug resistance of Aeromonas species in Ningxia was the most serious, and the antibiotic resistance trends were in line with the results of Zhang et al. [35]. Normally, Aeromonas spp. have inherent resistance to ampicillin, cefazolin, and amoxicillin/clavulanic acid. However, although the resistance rate of Aeromonas strains to ampicillin and amoxicillin/clavulanic acid reached more than 90% in this study, some Aeromonas strains were susceptible to ampicillin and amoxicillin/clavulanic acid, which is consistent with the results of Yano and Li [9,36]. The rate of drug resistance to cefazolin was low, which is consistent with the results of Wang Shuxian and Zhang Piao [37,38]. Fortunately, the rate of resistance of Aeromonas to other antibiotics is also low [1]. We found that the sensitivity of different species of Aeromonas spp. to antibiotics may differ, which indicates that we should be able to select drugs according to the species of Aeromonas which is causing infection [36]. The drug resistance spectrum of Aeromonas spp. might vary according to the geographical area. Five strains of Aeromonas (two strains of A. sobria were isolated from Qingtongxia Nature Reserve and three strains of A. hydrophila were isolated from Yellow River beach wetland) were isolated from the feces of migratory birds in Ningxia. The multidrug-resistant strains were resistant to nine kinds of antibiotics: gentamicin, cefazolin, cefotaxime, ampicillin, piperacillin, ampicillin-sulbactam, trimethoprim-sulfamethoxazole, chloroamphenicol, ciprofloxacin, and tetracycline. Although multidrug-resistant Aeromonas spp. have been widely reported [39], the emergence of multidrug resistance in Aeromonas spp. carried by migratory birds deserves more attention. Therefore, it is necessary continuously to monitor the emergence of determinants of antibiotic resistance and the drug resistance library [26]. The Aeromonas spp. carried by migratory birds mainly come from the environment. Therefore, it is urgent to evaluate the drug resistance of Aeromonas spp. in the environment. It is not only necessary to pay attention to the important role of migratory birds in the transmission of drug-resistant bacteria, but also to reduce the use of antibiotics in order fundamentally to reduce the transmission of drug-resistant bacteria.
The virulence genes were analyzed from the regional aspect: there were significant differences in the content of different virulence genes in the five regions. The relative abundance and diversity of virulence genes were the highest in Dali Lake; Ningxia was second. The comparison of virulence genes between the two regions in Inner Mongolia showed that although the diversity of virulence genes was the same, the differences in relative abundance were significant. It may be that a large number of migratory birds died in the Inner Mongolia region of China between 2017 and 2018; although Aeromonas spp. has not been isolated and identified, we cannot rule out that Aeromonas is one of the causes of disease in migratory birds [40]. From a comparative species perspective, A. hydrophila and A. veronii have high relative abundance and diversity of virulence. The number of strains containing three or more virulence genes in A. hydrophila and A. veronii were higher than in the other two species. This further confirmed that A. hydrophila and A. veronii are highly pathogenic Aeromonas spp. The virulence genes hlyA, aerA, alt, fla, and lip had very high detection rates. The rate of detection of ast was relatively low, and other virulence genes were not detected. This is consistent with the study of Li F, where hlyA, aerA, alt, fla, and lip were highly prevalent in A. hydrophila and A. veronii, but some genes such as laf and ascF-G were not common in all species [35]. It is worth noting that there were 25 virulence-gene combination patterns among the 176 isolates in our study, of which the aerA, aerA-lip, and hlyA-aerA-alt-fla-lip virulence-gene combination patterns were dominant. For the combination pattern of aerA and aerA-lip virulence genes, A. sobria was the main epidemic strain, while A. hydrophila was the main epidemic strain with the pattern hlyA-aerA-alt-flalip. Considerable differences in virulence-gene combination patterns may exist between different species. These differences may contribute to diversity in pathogenesis among different Aeromonas spp. [41]. Furthermore, the differential expression of virulence genes may be an important factor in the pathogenesis of Aeromonas spp. [6]. An experimental study showed a positive correlation between the number of virulence genes and the pathogenicity of Aeromonas spp. [42].
The ability of Aeromonas spp. to form biofilms could increase their drug resistance and the possibility of persistent infection [43]. The data of this study show that, under the same conditions, the ability of different species to form biofilm can be the same, and the ability of the same species to form biofilm can differ. In the study of Chenia and Duma, the ability of Aeromonas strains isolated from freshwater fish and seawater to form organisms is similar to the finding of this study [44]. In our study, 176 Aeromonas isolates were not strongly adherent, and most of them were only weakly adherent. These results are similar to the study of Isoken H Igbinosa [25]. Although recent research has shown that the fla gene is related to biofilm formation [43], no correlation was found in our study. This indicates that the complexity of biofilm formation of Aeromonas cannot be determined simply by the genotype.
Owing to the complexity of the classification of Aeromonas spp., many researchers have studied the phylogenetic tree of Aeromonas spp. using many different methods; 16S rDNA sequencing is the most common molecular tool in bacteriological classification, but it has proved to have several disadvantages. Housekeeping genes are considered to be suitable markers for phylogenetic analysis [45], and the phylogenetic tree can be more accurately constructed by sequence concatenation of six housekeeping genes [12]. In this study, a concatenated set of housekeeping genes (gltA-groL-gyrB-metG-PPSA-recA) was used to construct a phylogenetic tree by the neighbor-joining method. We evaluated the phylogenetic tree of 174 Aeromonas isolates based on the concatenated gene sequences (Figure 3), and revealed that these isolates were closely related and included A. sobria (107 isolates), A. hydrophila (53 isolates), A. veronii (9 isolates), and A. caviae (5 isolates). Although four Aeromonas species were identified according to the biochemical tests, there were only three large outgroups in the phylogenetic tree of housekeeper gene construction. This is because A. veronii is within the A. hydrophila cluster in three of the individual gene loci making up the MLST (gyrB, gltA, and metG) [46]. A. veronii is divided into A. veronii bv. Sobria and A. veronii bv. veronii, and some studies have shown that A. veronii bv. Sobria and A. sobria belong to the same taxon [12]. The genetic relationship between A. caviae and the other three strains is relatively distant. However, the use of several gene loci may mask the evolutionary history of individual genes, although an increase in the number of gene loci increases the resolution of the analysis by joining the combined capacities of all molecular clocks [46]. Between the phylogenetic tree of the PPSA and the concatenated phylogenetic tree, there are few differences in the overall clustering of most strains. This may be because the PPSA gene locus is the least variable. Some studies have shown similar results with the recA and the metG gene loci, which is consistent with this study ( Figure S2) [46]. This study verified that PPSA can be used to construct a more accurate phylogenetic tree. The results showed that the genetic distances in the phylogenetic tree constructed from housekeeping genes are related to the number of virulence genes (Figure 3, Figures S1 and S2). It was found that the phylogenetic tree constructed from housekeeping genes for A. sobria and A. hydrophila seems to be related to virulence genes. The relationship between A. sobria carrying multi-virulence and major A. hydrophila phylogroups was closer; A. hydrophila with few or no virulence genes is more closely related to major A. sobria phylogroups.

Conclusions
We obtained 176 Aeromonas isolates from migratory bird samples, and high genetic diversity was observed in these isolates. Virulence genes were examined by PCR, indicating that Aeromonas spp. Are well-equipped with potential virulence genes including hlyA, aerA, alt, ast, fla, and lip, and pose a risk to human health. When measuring antibiotic resistance to nine distinct antibiotic classes, 94.3% of the strains were found to be MAR (≥3). Our findings demonstrate that migratory birds can be reservoirs of virulent and multidrug-resistant Aeromonas, and act as a vehicle for the transfer of different genotypes of Aeromonas and antibiotic-resistant determinants to regions worldwide through migration. Strengthening the directed surveillance of wild migratory birds can effectively anticipate disease outbreaks, allowing for preventive or mitigating measures to be taken.

Supplementary Materials:
The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/microorganisms11010007/s1. Figure S1: Unrooted phylogenetic trees based on recA (A) and metG (B) gene sequences, showing relationships in the genus Aeromonas from migratory birds in this study; Figure S2: Unrooted phylogenetic trees based on groL (A), gltA (B) and gyrB (C) gene sequences, showing relationships in the genus Aeromonas from migratory birds in this study; Table S1: 21 kinds of antibiotics in BD Phoenix TM-100 automatic microbial identification system; Table S2: In vitro susceptibility of 176 Aeromonas isolates to 13 antimicrobial agents; Table  S3: Antibiotic susceptibility test (AST) typing of Aeromonas spp.; Table S4: Genetic detection of six virulence genes in Aeromonas spp.
Author Contributions: B.L. and X.J. designed the experiments and drafted the manuscript. B.J. and T.Y. processed the samples. C.L.M.G., T.W. and Y.L. collected the data and samples. L.Z. contributed to the statistical analyses. J.L. supervised the overall work. X.G. discussed the results. Y.S. obtained funding and coordinated the project, and revised the manuscript. All co-authors contributed to the article and approved the submitted version. All authors have read and agreed to the published version of the manuscript.
Funding: This research was funded by the National Science and Technology Major Project of China, grant number 2018ZX10733402.

Data Availability Statement:
The data that support the findings of this study are available from the corresponding author upon reasonable request.

Conflicts of Interest:
The authors declare no conflict of interest.