Metallic nanoparticles and ions accelerate the uptake of extracellular antibiotic resistance genes through transformation

Background: Antibiotic resistance genes (ARGs), heavy metal ions and nanoparticles (NPs) are emerging and ubiquitous contaminants in the environment. However, little is known about whether heavy metal-based NPs or ions could facilitate the dissemination of ARGs through natural transformation. This study evaluated the contributions of heavy metal-based NPs (Ag NPs, CuO NPs and ZnO NPs) and their ion forms (Ag + , Cu 2+ and Zn 2+ ) to the transformation of extracellular ARGs in Acinetobacter baylyi ADP1. Results: We found that these commonly-used NPs and ions from environmentally relevant concentrations can significantly promote the natural transformation frequency of ARGs by a factor of 11.0-folds, which is comparable to the effects of antibiotics. The enhanced transformation by Ag NPs, CuO NPs, Ag + and Cu 2+ was primarily associated with reactive oxygen species (ROS) overproduction and cell membrane damage, which was also evident from up-regulations of both transcription and translation of ROS and outer membrane-related genes. Additionally, transmission electron microscope imaging revealed the roughened cell membrane after Ag NPs, CuO NPs, Ag + and Cu 2+ exposure. ZnO NPs and Zn 2+ might increase the natural transformation rate by stimulating the stress response and ATP synthesis. All tested NPs and ions resulted in up-regulating the competence and SOS response-associated genes. Conclusions: Our results demonstrate that Ag, CuO and ZnO-based NPs/ions from environmental concentrations could promote the natural transformation of plasmid-encoded ARGs into naturally competent A. baylyi . Our findings provide insights into the contributions of heavy metals and NPs to the spread of antibiotic resistance. B, Li JW. Nanoalumina promotes the horizontal transfer of multiresistance


Background
The dissemination of antimicrobial resistance (AMR) is posing a progressing global health crisis.
Disquietingly, based on the scenario analysis, the annual deaths induced by AMR-provoked infection are predicted to reach 10 million by 2050 [1]. One of the primary drivers responsible for the increasing prevalence of AMR is the horizontal gene transfer (HGT) of AMR among various bacteria, which is consisted of three mechanisms: i) Conjugation: the dissemination of antibiotic resistance genes (ARGs) from a donor cell to a recipient cell via pilus-driven physical contact; ii) Transformation: recombination of foreign ARGs after bacterial uptake of exogenous genetic materials; and iii) Transduction: phage-mediated DNA transfer upon infection [2].
In particular, transformation can take place in more than 80 naturally competent bacteria with distant phylogenetical backgrounds, even consisting of human pathogens [3] Although these bacteria share a wide phylogenetic distribution, the key steps involved in transformation among these species are similar, which consist of: capture of exogenous double-stranded DNA (dsDNA) via type IV transformation pilus or type II secretion systems, degradation of dsDNA into single-stranded DNA (ssDNA), internalization of ssDNA via a DNA translocase complex at the cytoplasmic membrane and lastly, the recombination of exogenous DNA after the homology search [4,5]. In the clinical aspect, antibiotic stressors such as aminoglycoside and fluoroquinolone could induce the competence for natural transformation in the human pathogen (e.g., Streptococcus pneumoniae and Legionella pneumophila) [6,7]. Worryingly, due to the prevalence of extracellular ARGs, antibiotics residual and naturally competent bacteria in the environment, the environmental transformation of ARGs was estimated to be quite frequent, but which has been largely overlooked [8][9][10].
Basically, free-living extracellular ARGs, heavy metal-based NPs and ions ubiquitously co-exist in the same environment (e.g., wastewater treatment plants (WWTPs)) [22,23]. Thus, it is of significance to evaluate whether the co-existence of heavy metal-based NPs/ions and ARGs could promote the 4 dissemination of ARGs via natural transformation.
In this study, we aim to investigate whether heavy metal-based NPs and ions could promote the dissemination of ARGs via natural transformation. A transformation model was established by using pWH1266 plasmid carrying bla TEM-1 and tetA as the exogenous ARGs, and A. baylyi ADP1 as the recipient, to investigate the effects of heavy metal-based NPs (including Ag NPs, CuO NPs and ZnO NPs) and their ion forms (Ag + , Cu 2+ and Zn 2+ ) on natural transformation. The underlying mechanisms were investigated by detecting the oxidative stress, cell membrane permeability and transmission electron microscopy (TEM) imaging, in conjunction with genome-wide RNA sequencing and proteomic analyses.

Results
Heavy metal-based NPs and ions increased transformation frequency A naturally competent opportunistic pathogen A. baylyi was used to evaluate the effects of heavy metal-based NPs (including Ag NPs, CuO NPs and ZnO NPs) and their ionic forms (Ag + , Cu 2+ and Zn 2+ ) on the transformation of plasmid pWH1266 encoded ARGs. The tested heavy metal and NP concentrations were included the environmentally relevant concentrations (e.g., 0.1 and 1 mg/L). In general, all the tested heavy metal-based NPs and ions could significantly (* p < 0.05, ** p < 0.01) promote the transformation of pWH1266 plasmid into A. baylyi at certain exposure levels (Fig. 1A, B and C). For example, the transformation frequencies under 0.1 mg/L Ag + (7.5 ± 1.1 × 10 − 6 per recipient cell, Fig. 1A), 100 mg/L CuO NPs (1.5 ± 0.04 × 10 − 5 per recipient cell, Fig. 1B) and 10 mg/L Zn 2+ -treated group (2.9 ± 0.2 × 10 − 5 per recipient cell, Fig. 1C) were 2.8, 5.6 and 11.0-folds higher than that of the control groups (2.7 ± 0.5 × 10 − 6 per recipient cell), respectively. Various NPs and ion types resulted in different trends in terms of transformation frequencies. For Ag and CuO NPs/ionstreated groups, the increments of the natural transformation frequency were concentrationindependent when compared to the control groups. For the ZnO NPs/ions-treated groups, a concentration-dependent increase was observed for the natural transformation frequencies when compared to the control groups. From 0.1 mg/L to 100 mg/L, the ZnO NPs/ions-mediated 5 transformation frequencies increased with the increments of ZnO NPs/ions concentrations (Fig. 1C).
Multiple approaches were conducted to confirm the uptake of plasmid pWH1266 by A. baylyi. Firstly, the minimum inhibitory concentrations (MICs) of recipient wild-type A. baylyi and transformants against both Amp and Tet antibiotics were scanned. Since the plasmid pWH1266 encodes resistance genes against ampicillin (Amp) and tetracycline (Tet), transformants should have obtained the resistance against Amp and Tet from the plasmid. As expected, all the transformants exhibited around 15 and 5 folds higher MICs to Amp and Tet, compared to the recipient (Fig. 1D). Secondly, plasmids extracted from the transformants were compared with the donor plasmid by gel electrophoresis. Clear bands from transformants were shown with approximate sizes to the donor ( Fig. 1E), indicating that the transformants have received pWH1266 plasmids. Thirdly, the Polymerase chain reaction (PCR) amplification with long amplicons was employed to confirm if the transformants have carried both bla TEM−1 and tetA genes. All the amplified products from transformants exhibited similar sizes to that of the donor (Fig. 1E), demonstrating that the transformants have harbored plasmid pWH1266.
Collectively, these results confirmed that heavy metal-based NPs and ions could promote the transformation of pWH1266 plasmid into A. baylyi at environmentally relevant concentrations.
ROS over-production under the exposure of heavy metal-based nanoparticles and ions All of the tested Ag, CuO and ZnO NPs/ions were able to increase the intracellular ROS generation of A. baylyi. Based on flow cytometer detection results, the recipient strain showed significant (* p < 0.05, ** p < 0.01) concentration-dependent increases of intracellular ROS generation from 0.01 mg/L Ag NPs and CuO NPs, and from 0.1 mg/L Ag + and Cu 2+ , to 100 mg/L Ag and CuO NPs/ions ( Fig. 2A and   B). More importantly, it was found that the Ag and CuO NPs/ions-increased transformation was correlated with the increase of ROS levels below a certain threshold (approximately around 2-fold of 6 ROS increase compared to the control). For instance, the transformation frequencies of pWH1266 plasmid started to decrease from 10 mg/L of Ag + , Ag NPs and CuO NPs ( Fig. 1A and B), at which the ROS generation mediated by the corresponding heavy metal-based NPs and ions was over 2-fold higher than the control groups ( Fig. 2A and B). In contrast, the ZnO NPs/ions-facilitated transformation frequencies did not correlate with the ROS generation mediated by ZnO NPs/ions ( Fig. 1C and 2C).
Moreover, although the ROS generation increased under the exposure of ZnO NPs/ions from 0.01 mg/L, the fold-changes were below 1.3-fold (Fig. 2C), which is much lower than the increment of ROS mediated by Ag and CuO NPs/ions (e.g., 2.7-fold increase under 100 mg/L Ag + , Fig. 2A).
The molecular responses of A. baylyi against heavy metal-based NPs/ions were further investigated by using whole-genome RNA sequencing and proteomic analysis. In terms of ROS response, under all heavy metal-based NPs/ions treatments, the expression levels of the 12 known antioxidant systemrelated genes were mostly up-regulated, especially for the expression of the alkyl hydroperoxide reductase-coding genes ahpC and ahpF (Fig. 2D, Table S2). To illustrate, in response to the exposures of 1 mg/L Cu 2+ , the expression level of ahpC gene were 42.8-fold (i.e., Log2 fold change (LFC) = 5.42) higher than that of the control. To further validate the oxidative stress response in translational levels, proteomic sequencing was performed and indicated a similar enhancement of the antioxidant system ( Fig. 2E, Table S3). Apart from the increased translations of alkyl hydroperoxide reductase AhpC and AhpF, there is also an upregulation in the translation of catalase KatA under all heavy metal-based NPs/ions, in which up to 7.1-fold (i.e., LFC = 2.8) increase was observed when treated with 10 mg/L Zn 2+ (Fig. 2E).
To further verify whether heavy metal-based NPs and ions-mediated ARGs transformations were correlated to the ROS over-production, we then examined the effect of a ROS scavenger, thiourea, on ROS production and ARGs transformation. With 100 µM thiourea added, the ROS production levels were significantly (* p < 0.05, ** p < 0.01) reduced to the extent of control groups across most of the heavy metal-based NPs and ions dosage (Fig. 2F). Correspondingly, the Ag and CuO NPs/ionsmediated ARGs transformation frequencies were significantly (* p < 0.05, ** p < 0.01) decreased to the extent of control groups after thiourea addition, while thiourea did not reduce the ZnO NPs/ionsmediated ARGs transformation to the extent of control groups (Fig. 2G). These results further validated the correlation between ROS over-production and ARGs transformation under the exposure of Ag and CuO NPs/ions. Heavy metal-based nanoparticles and ions increased cell membrane permeability Cell membrane permeability of heavy metal-based NPs and ions-treated recipient was also evaluated by flow cytometer to verify whether it was associated with transformation enhanced by heavy metalbased NPs and ions (Fig. 3). Similar to the ROS generation, the concentration-dependent increases of membrane permeability could also be detected within Ag and CuO NPs/ions treated A. baylyi ( Fig. 3A and B). In contrast, there was no significant change across all ZnO NPs/ions-treated A. baylyi (Fig. 3C).
TEM imaging of A. baylyi was also conducted to evaluate the changes in cell membrane morphology under different concentrations of heavy metal-based NPs and ions treatment (Fig. 3D). Compared to the control groups, the increased cell membrane roughness and leakage of cytoplasm could be observed under 1 mg/L Ag + , Cu 2+ and CuO NPs dosage. In comparison, no obvious changes in cell membrane morphology could be observed under 10 mg/L ZnO NPs/ions (Fig. 3D).
In terms of RNA transcription, most of the outer membrane-related genes (e.g., 20 out of 23 genes under 1 mg/L Ag + treatment) were only moderately altered after 2 h of all heavy metal-based NPs and ions treatment (0 ≤ |LFC| ≤ 1). As exceptions, the outer membrane assembly gene bamD were 1.1 to 1.3-fold (LFC) up-regulated under Ag and CuO NPs/ions treatment, and the ACIAD0121 and adeK genes were down to -1.49 to -2.54-fold down-regulated under all heavy metal-based NPs and ions treatment (Fig. 3E, Table S4). Differently, the translation of outer membrane-related protein  Table S5).
Heavy metal-based NPs and ions stimulated transcription and translation of competence, stress response, SOS response and ATP production-related genes The key steps involved in transformation of A. baylyi ADP1 are consist of type IV transformation pilus systems (pil gene family), translocase complex at the cytoplasmic membrane (com gene family) and recombination of exogenous DNA after the homology search [4,5]. The heavy metal-based NPs/ionsmediated transcription and translation response of the genes involved in the competence system were further evaluated (Fig. 4A). Firstly, for competence-related genes, three genes associated with type IV transformation pilus and DNA translocase systems (comEA, pilT and pilU) showed increased transcription (e.g., LFC = 1.  (Fig. 4B). Thirdly, we found that heavy metal-based NPs/ions dosage elevated the transcription of the majority (e.g., 28 out of 34 genes under 1 mg/L Ag + treatment)of SOS response-associated genes (Fig. 4B) and the translation of four SOS response-associated proteins (DnaN, HupB, RecR and Ssb, Fig. 4C). More obviously, ZnO NPs/ions considerably enhanced the transcription of the stress response genes (Fig. 4B), in which the transcription of nirD genes were 9.6-fold (LFC = 3.3) higher than the control group when treated with Zn 2+ . Lastly, although the transcriptions of ATP-related genes were not up-regulated (Fig. 4B), the translations of those ATP-related proteins were largely promoted under CuO NPs/ions and Zn 2+ treatment (Fig. 4C).

Heavy metal nanoparticles and ions promoted ARGs transformation at environmentally relevant concentrations
In this study, we observed that heavy metal-based NPs/ions exposure could boost the natural transformation phenotype in A. baylyi at certain concentrations. Under Ag NPs/ions and Cu 2+ treatment, the transformation frequencies increased at low concentrations (e.g., 0.01 to 1 mg/L Ag NPs), but decreased above a certain threshold (e.g., 10 mg/L Ag + ), due to the decrease of transformants and recipient numbers caused by bactericidal effect from the corresponding NPs/ion. In addition, the enhanced transformation is concentration-dependent for ZnO NPs and Zn 2+ exposure.
The successful transformation of pWH1266 was confirmed using multiple approaches. Our results suggest that heavy metal-based NPs/ions at environmentally relevant concentrations (e.g., 0.01 mg/L CuO NPs/ions) could promote the natural transformation of ARGs to naturally competent A. baylyi.
This study employed multiple approaches to elucidate the underlying mechanisms, including flow cytometry to measure membrane permeability and ROS generation, gene expression analysis by whole-genome RNA sequencing, and quantitative proteomic response analysis. Based on the phenotypic and genotypic data, we proposed the mechanisms underlying heavy metal NPs/ionspromoted natural transformation ( Figure 5).
The heavy metal NPs/ions-promoted natural transformation was found to be associated with ROS overproduction, as validated by the reversal of the transformation frequency to baseline levels with the addition of a ROS scavenger-thiourea ( Figure 2G), with ZnO NPs as an exception. The observed transformation promoted by heavy metal-based NPs/ions was in agreement with our previous findings that CuO and Ag NPs/ions could enhance conjugative ARGs transfer from Escherichia coli to Pseudomonas putida via over-generation of ROS (18,19) Likewise, recent studies have also indicated that solar disinfection and water disinfection by-products could increase the natural transformation rates of extracellular DNA in A. baylyi via inducing ROS (24,25).
In addition, we also observed that increased cell membrane permeability was associated with the CuO and Ag NPs/ions-enhanced transformation efficiency (Figure 3). This damaged cell membrane integrity might lead to the formation of cell membrane channels, and enhance the function of the secretion competence systems for an easier ARGs uptake (26,27). Indeed, we observed up-regulated transcription (comEA, pilU and pilT, Figure 4A promoting the plasmid uptake of Acinetobacter baumannii (35). Moreover, heavy metal-based NPs/ions were reported to trigger the co-selection of AMR (12,36), directly induce AMR via mutagenesis (37,38), as well as facilitate the lateral transfer of ARGs via conjugation (18)(19)(20).
Considering the co-occurrence of heavy metal-based NPs/ions and ARGs contamination in the environment, heavy metal-based NPs and ions might pose a substantial risk of AMR dissemination among environmental microbial via co-selection, mutagenesis, conjugation and natural genetic transformation.

Conclusions
In summary, our results demonstrate that Ag, CuO and ZnO-based NPs/ions from environmental concentrations promote the natural transformation of plasmid-encoded ARGs by naturally competent A. baylyi, which is comparable to the effect of antibiotics. The Ag and CuO NPs/ions-induced enhancements of natural transformation was associated with the ROS over-production and cell membrane damage, which could be prevented by the addition of a ROS scavenger. Contrarily, ZnO NPs/ions could increase the natural transformation through provoking stress response and ATP synthesis. Conclusively, all tested NPs and ions might promote the natural transformation of ARGs by up-regulating the competence and SOS response-associated genes. More in situ assessment on the potential risk of heavy metal-based NPs and ions mediated horizontal transfer of ARGs is recommended.

Bacterial strains, culture media and nanoparticles
In this study, plasmid pWH1266 (8.89 kbps, ATCC ® 77092™) carrying two ARGs, tetA against Tet and

Measurements of MICs, ROS, and cell membrane permeability
In order to confirm the successful transformation, the MICs of A. baylyi recipient and transformants against Amp and Tet were measured. To investigate the effects of ROS and cell membrane integrity on the transformation frequencies, both ROS production and cell membrane permeability of the A. baylyi were measured by a CytoFLEX flow cytometer (Beckman Coulter, USA) based on the previous procedure [34]. Each experiment was conducted with biological triplicates. USA) at 80 kV according to the method previously described [18].

RNA extraction, genome-wide RNA sequencing and bioinformatics
The transformation systems were established as described above, with the dosages of 0 mg/L (control), 1 mg/L Ag NPs or Ag + , 1 mg/L CuO NPs or Cu 2+ and 10 mg/L ZnO NPs or Zn 2+ . The total RNA was extracted by the RNeasy Mini Kit (QIAGEN®, Germany) after a 2-h mating period. In total 21 RNA samples from the control and heavy metal-based NPs/ions-treated groups (3 biological samples from each of the 7 groups) were performed for strand-specific cDNA library construction and Illumina paired-end sequencing (HiSeq 2500, Illumina Inc., San Diego, CA) at Macrogen (Korea), which generated around 800 Mbp data for each sample. The bioinformatics pipeline was reported in our previous study [34] and described in the supporting information. Differences in gene transcriptional values were calculated between untreated and heavy metal NPs/ions-treated group by determining the LFC of the averaged fragments per kilobase of a gene per million mapped reads (FPKM) values.

Protein extraction and proteomic analysis
Another set of transformation systems the same as the RNA extraction treatment described above was set, only the mating time extent to 6 h. Similarly, 21 bacterial samples were harvested by at 12,000 × g centrifugation for 10 min. Protein was extracted using the B-PER method, and protein data were analyzed by the ProteinPilot software (ABSciex, USA), the R-based program Msstats, and PeakView v2.1 (ABSciex, USA). The detailed preteomic sequencing and bioinformatics procedures were reported in our previous study [34] and described in the supporting information.

Statistical analysis and data availability
Data analysis was used by SPSS 19.0 (IBM, Armonk, USA). Significant differences were performed by Independent-samples t-test. A value of p < 0.05 was considered significant. Data were expressed as mean ± standard deviation. All sequencing data in this study have been deposited in publicly accessible databases. RNA sequence data were accessible through Gene Expression Omnibus of NCBI (accession no. GSE139295). The mass spectrometry proteomics data were deposited to the ProteomeXchange Consortium via the PRIDE partner repository (PXD012641).