To protect human health, it is necessary to understand the distribution and transfer risk of ARGs. To date, research has concentrated on the geographical distribution of ARGs (Liu et al. 2018a, Yang et al. 2019). In the present study, three important rivers in northeast China, the Heilongjiang River, Tumen River, and Yalu River, were assessed. River water samples and riverbank soil samples were collected to analyze the ARG composition using metagenetic methods. The relative abundance of ARGs in river water was between 7.5 × 10− 1 and 1.25 ×100 per 16s rRNA gene, and that in the riverbank soil was between 1.75 × 10− 1 and 2.25 ×10− 1 per 16s rRNA gene. The abundance of resistance genes in rivers and soil were similar to the range of ARG abundances reported previously (Li et al. 2015). In terms of composition, multidrug resistance genes represented about half of the total abundance, which was related to the high expression of efflux pumps (Hiroshi and Jean-Marie 2012). Other types, such as bacitracin, beta-lactam, and tetracycline resistance genes, also showed high abundance. As expected, these dominant ARG types corresponded to the antibiotics that have been used widely for human treatment and livestock breeding. BacA is a ubiquitous ancient resistance gene, which has been obtained by many bacteria during the process of evolution, resulting in the high abundance of bacitracin resistance genes. Among the multidrug resistance genes, multidrug efflux pumps are ubiquitous in bacteria, which not only effectively reduce the concentration of antibiotics, but also participate in other processes, such as detoxification of metabolic intermediates, virulence, and signal trafficking (Bao et al. 2016). The presence of a variety of ARGs with efflux function, like adeF and mexF, lead to a high proportion of multidrug resistance genes. The increase of polymyxin resistance genes is a concern. Polymyxin is termed the last line of defense of human antibiotics treatment, and the increase in polymyxin resistance genes represents a potential threat to human health.
Among the samples from the three rivers, the abundance of ARGs was highest in YLJ river water and soil, and was lowest in TMJ. In previous studies, the abundance and composition of ARGs in water was seriously affected by human activities, which might be one of the main driving forces for the formation of ARG structures in water. The population living along the coast of YLJ is significantly larger than that of TMJ and HLJ, and the urbanization process has been relatively faster, which might be one of the reasons for the high abundance of resistance genes in YLJ. In terms of their composition, ARGs in soil had certain characteristics. Vancomycin resistance genes accounted for a large proportion in soil, but not in river water. Vancomycin is used to treat humans and was originally produced by Amycolatopsis orientalis, which was isolated from soil. There are more vancomycin-producing bacteria in soil, thus the abundance of vancomycin resistance genes was very high. Except for vancomycin resistance genes, the composition of the ARGs in soil was very similar to that in river water, and almost all the resistance gene types detected in soil were found in river water. This might be related to the characteristics of the river itself. When the river is in flood, the water level rises, covering part of the riverbank. In this process, the bacteria and resistance genes in the soil and river water are mixed and exchanged. In the dry season, the water level of the river drops and the riverbank is exposed, which reduces the mixing and exchange of bacteria and resistance genes. The samples in this study were collected in September, which is in the dry season of these rivers. The composition and structure of the bacteria and resistance genes in river water and soil are relatively stable in this period, which might explain the similar composition of ARGs detected in this study. This similarity suggested that rivers could obtain ARGs from the riverbank soil, and the soil could act as a reservoir of ARGs, resulting in secondary ARG pollution.
Notably, the composition of ARGs in the three rivers was very similar, regardless of types or subtypes. The Heilongjiang River, Tumen River, and Yalu River flow in different directions and are located in three different provinces. Although TMJ and YLJ come from the same origin, there are great differences in population density and economic development in the regions they flow through. The results of ecological boundary analysis showed that the abundance and diversity of ARGs did not have a clear demarcation with the change of longitude and latitude. We next considered whether the similarity in ARGs was determined by climate conditions. These three rivers are all located in northeast China with a northern temperate climate, relatively low temperature, and similar precipitation and other conditions. Therefore, we hypothesized that climate played a decisive role in the composition of ARGs. Some ARG data in different niches were reported in previous studies. In four sewage treatment plants in Harbin, tetracycline and sulfonamide resistance genes were detected (Wen et al. 2016). These ARG subtypes were also detected in river water samples with high abundance in this study. In a study on the effect of soil fertilization on ARGs in northeast China (Li et al. 2020), the reported highly abundant ARGs types were also detected as highly abundant in the present study. In particular, the ARG data in Heihe (HH) were similar to our data. In a large-scale survey of ARGs in drinking water (Ma et al. 2017), bacitracin, multidrug, and sulfonamides resistance genes were the top three ARG types in Heilongjiang Province. These results suggested that the composition of ARGs in different niches in northeast China is similar, indicating that climate might be the decisive factor in the occurrence and distribution of ARGs. Currently, there are no large-scale sequencing data on the distribution of ARGs in the natural environment of northeast China, whether in water, soil, or other niches. Therefore, this conjecture needs to be further verified in future studies.
Compared with ARGs in Yangtze River, ARGs abundance in the three rivers was lower, especially super ARGs like mcr-1 and NDM-1. According to a previous study, mcr-1 and NDM-1 were about 1.2×109copies/L and 4.1×106 copies/L in the downstream of Yangtze River(Wang et al. 2019). In contrast, the highest abundance of mcr-1 was only 4.0 copies/L and NDM-1 was not detected in the three rivers. For common ARGs, the abundance of 16s rRNA gene, suls(sul1 and sul2), tets(tetA, tetC, tetE, tetO, and tetW), erms(ermB and ermF), and qnrs(qnrB and qnrS) ranged from 2.9×108 to 2.4×1010 copies/L, 1.1×107 to 3.6×108 copies/L, 1.5× 106 to 4.9×107 copies/L, 8.2×105 to 1.8×107 copies/L, and 7.7×103 to 1.5×106 copies/L, respectively, with their mean concentration were 1.9×108, 2.0×107, 6.8×106, and 4.2×105 copies/L, correspondingly(Yang et al. 2020, Zhang et al. 2020a). In this study, these ARGs abundance ranged from 2.1×10− 3 to 3.6×10− 2 copies/16s rRNA, 7.6×10− 4 to 4.4×10− 2 copies/16s rRNA, 0 to 4.9×10− 3 copies/16s rRNA, 1.8×10− 4 to 7.5×10− 2 copies/16s rRNA. It seems that ARGs in the three rivers were less than that in the Yangtze River. Actually, there were more ARG subtypes detected in river water samples like sul3 and sul4. In result, sulfonamide resistance genes were more than suls, so was others.
To further evaluate the potential threat caused by ARGs, the hosts carrying ARGs were analyzed, especially human pathogenic bacteria. Pseudomonas carry many ARGs. Pseudomonas is widely distributed in nature and is closely related to human infection, for example, Pseudomonas aeruginosa, Pseudomonas fluorescens, and Pseudomonas putida. Among Pseudomonas species, Pseudomonas aeruginosa was the most common and carried more ARGs, among which multidrug resistance genes accounted for 60% of the total. In a previous study, 82% of Pseudomonas aeruginosa isolated from hospital sewage treatment plants showed resistance to multiple antibiotics. Pseudomonas aeruginosa can cause serious infections, and is very common in nosocomial infections. The presence of different ARGs, especially multidrug resistance genes, makes treatment challenging. Among all samples in this study, Escherichia coli was the pathogen that carried the most ARGs. Worldwide, antibiotic-resistant E. coli are widely present in all kinds of niches. Third-generation cephalosporin-resistant E. coli was detected in coastal surface waters in England and Wales (Leonard et al. 2015). In Dutch rivers, 27.6% of E. coli are resistant to at least one antibiotic (Blaak et al. 2014) and 42% of E. coli in the river Seine in France have at least one antibiotic resistance gene (Servais and Passerat 2009). Other antibiotic resistant pathogenic bacteria are related to human diarrhea, such as Aeromonas caviae and Bacteroides fragilis, and to respiratory infection, like Bordetella bronchiseptica. A previous study has shown that humans can acquire antibiotic resistant bacteria from water while performing water sports or other activities (Leonard et al. 2015). The three rivers in this study were closely related to the daily lives of residents; therefore, we should be vigilant concerning the related risk of infection.
To further understand risk of ARG diffusion, it is important to evaluate the potential horizontal transfer of ARGs. MGEs participate in this process directly. The ARGbacteria network showed the potential hosts of ARGs. In the network, Aeromonas was predicted to correlate positively with many types ARGs, including beta-lactam, multidrug, tetracycline, and MLS type genes, and experimentally, these ARGs were detected in Aeromonas. This showed that the network was an effective way to find potential hosts of ARGs. Positively correlations between different types of ARGs, such as that between the multidrug resistance gene ompF and the beta-lactam resistance gene PBP-1B, indicated their co-occurrence. Different ARG subtypes seem to be preferred by different hosts. The multidrug resistance gene ompF correlated negatively with the Proteobacteria genus Rhodoferax, but positively with the genus Trabulsiella. In the ARG-MGEs network, many MGEs correlated positively with ARGs, which is consistent with previous studies (Zhao et al. 2019, Zheng et al. 2018). Some ARG subtypes correlated positively with many MGEs, such as OXA-12 and mdtH, indicating a high risk of horizontal transfer.
There are some limitations to this study. In the process of collecting soil samples, only one mixed sample was collected from each sample site. To improve the representativeness of the samples, soil samples were collected every 1 m and mixed for analysis. For groups with fewer sample sites, there might be a great difference between a single sample data and the average, thus a more intensive sampling strategy might be considered in future research. Moreover, the impact of season should be considered. The sampling time of this study was autumn, and the high temperature in summer might promote the growth and reproduction of many kinds of eukaryotes and strongly affect the distribution of bacteria and ARGs. Therefore, our results are only strictly applicable to the time-frame in which the samples were collected, and different results might be obtained if the study was repeated in other seasons.