Genome-Wide Bacterial Toxicity Screening ... - ACS Publications

Article pubs.acs.org/est

Genome-Wide Bacterial Toxicity Screening Uncovers the Mechanisms of Toxicity of a Cationic Polystyrene Nanomaterial Angela Ivask,†,‡,¶ Elizabeth Suarez,†,‡ Trina Patel,†,§ David Boren,†,§ Zhaoxia Ji,†,‡ Patricia Holden,†,∥ Donatello Telesca,†,§ Robert Damoiseaux,†,‡ Kenneth A. Bradley,†,‡,⊥ and Hilary Godwin*,†,▽ †

University of California Center for Environmental Implications of Nanotechnology, University of California, Los Angeles, California 90095, United States ‡ California NanoSystems Institute, University of California, Los Angeles, California 90095, United States ¶ Laboratory of Molecular Genetics, National Institute of Chemical Physics and Biophysics, Tallinn, Estonia § Department of Biostatistics, School of Public Health, University of California, Los Angeles, California 90095, United States ∥ Donald Bren School of Environmental Science and Management, University of California, Santa Barbara, California 93106, United States ⊥ Department of Microbiology, Immunology, and Molecular Genetics, University of California, Los Angeles, California 90095, United States # Department of Environmental Health Sciences, UCLA School of Public Health, University of California, Los Angeles, California 90095, United States ▽ Institute of the Environment and Sustainability, University of California, Los Angeles, California 90095, United States S Supporting Information *

ABSTRACT: By exploiting a genome-wide collection of bacterial single-gene deletion mutants, we have studied the toxicological pathways of a 60-nm cationic (amino-functionalized) polystyrene nanomaterial (PS-NH2) in bacterial cells. The IC50 of commercially available 60 nm PS-NH2 was determined to be 158 μg/mL, the IC5 is 108 μg/mL, and the IC90 is 190 μg/mL for the parent E. coli strain of the gene deletion library. Over 4000 single nonessential gene deletion mutants of Escherichia coli were screened for the growth phenotype of each strain in the presence and absence of PSNH2. This revealed that genes clusters in the lipopolysaccharide biosynthetic pathway, outer membrane transport channels, ubiquinone biosynthetic pathways, flagellar movement, and DNA repair systems are all important to how this organism responds to cationic nanomaterials. These results, coupled with those from confirmatory assays described herein, suggest that the primary mechanisms of toxicity of the 60-nm PS-NH2 nanomaterial in E. coli are destabilization of the outer membrane and production of reactive oxygen species. The methodology reported herein should prove generally useful for identifying pathways that are involved in how cells respond to a broad range of nanomaterials and for determining the mechanisms of cellular toxicity of different types of nanomaterials.

■

INTRODUCTION To satisfy the needs of the growing field of nanotechnology, production of engineered nanomaterials (NMs) with novel physicochemical properties is exponentially increasing. Safe incorporation of NMs into consumer products requires an understanding of their toxicology. Although NMs are being increasingly produced and incorporated into consumer products, there are growing concerns about the unintended release of these materials into the environment and potential hazardous effects on ecological systems.1 Ideally, we would like to be able to produce nanomaterials that are effective for their desired applications but do not exhibit undesirable or unnecessary toxic effects. To achieve this goal, methods and studies are needed that elucidate how physicochemical © 2011 American Chemical Society

properties of NMs correlate with both activity and toxicity, and ideally reveal why different physicochemical properties elicit specific biological responses (i.e., the mechanism of toxicity for specific NMs). Polystyrene nanomaterials (PS NMs) have been widely produced and are commercially available with different surface funtionalities and surface charge in a series of sizes. Critically, PS NMs have been included by the Working Party on Manufactured Nanomaterials in a list of manufactured Received: Revised: Accepted: Published: 2398

September 15, 2011 November 26, 2011 December 7, 2011 December 7, 2011 dx.doi.org/10.1021/es203087m | Environ. Sci. Technol. 2012, 46, 2398−2405

Environmental Science & Technology

Article

pathways of different nanomaterials. As such, this methodology should also prove useful for determining the mechanisms of action of a broad range of toxins and therapeutic agents. In combination with transcriptome level assays like the one recently proposed by Guo and Gu,13 this method constitutes an important tool in mechanistic toxicology more generally.

nanoscale materials for which environmental health and safety testing is a priority (http://www.epa.gov/opptintr/nano/). Systematic investigation of positively and negatively charged and neutral PS NMs in a broad range of eukaryotic cell lines and organisms has consistently revealed that positively charged, amino-functionalized PS-NH2 NMs are more toxic than the neutral and negatively charged (PS-COOH) analogs and that 60-nm PS-NH2 NM is more toxic than larger PS-NH2 NMs.2 Similar findings showing high toxicity of cationic NMs compared to neutral or negatively charged analogs have also been demonstrated for silicon NMs in mammalian cells.3 Targeted assays designed to probe the toxicity of cationic NMs suggest that, at least in eukaryotic cells, the mechanism of toxicity of these cationic materials involves ATP depletion and production of reactive oxygen species (ROS) that result from the damage of mitochondrial inner membrane.2,3 Whether or not similar mechanisms of toxicity of cationic nanomaterials also hold true in prokaryotes has not been reported. In addition, whether nanotoxicology researchers have seen ROS in their studies because that is what they were looking for, whether the ROS that is observed is a primary route of cell injury for NMs, or whether the observed ROS is the byproduct of other toxicological pathways have persisted an important unresolved questions. For example, Lyon and Alvarez have demonstrated that a different carbonaceous NM (C60) exerts its toxicity in bacteria through ROS-independent protein oxidation.4 To address these issues, we used a genome-wide toxicity assay that involves the characterization of growth phenotypes of a collection of E. coli nonessential single-gene deletion mutant strains (KEIO collection of ∼4000 mutants5) in the presence of a cationic polystyrene nanomaterial. A similar approach has been utilized previously to find the relative fitness of gene deletion strains of bacterial (E. coli) or yeast (S. cerevisiae) cells upon exposure to small molecules or environmental stressors.6−11 Here, the assay was optimized and validated to be applicable for quantitative high-throughput toxicological analysis for nanomaterials. Specifically, the assay was performed in liquid media as opposed to agarized conditions used in earlier studies because6−11 NMs often interact with cells via surface contact/attachment which is only possible under liquid conditions. Due to inherent properties of nanomaterials, such as their tendency to agglomerate or transform under biological conditions,12 optimization of the assay included careful selection of NM dispersion conditions. The results of this E. coli genome-wide toxicity assay revealed that the primary mechanisms of toxicity of cationic nanomaterial (60-nm PSNH2) in prokaryotic cells are (i) destabilization of the outer membrane (with LPS and several transmembrane proteins playing a critical role in stabilizing the outer membrane), and (ii) ROS generation, both intracellularly and at the cell membrane (with ubiquinone playing a critical role in mitigating this effect either through direct interaction with ROS and/or due to the role that ubiquinone plays in normal ETC activity). A similar toxicity profile was shown for a small molecule with a highly positive surface charge (cationic peptide polymyxin B) indicating that the observed toxicity pathways were due to the positive surface charge of the cationic nanomaterial. The methodology described herein provides an important tool for identifying toxicologial pathways for nanomaterials. A particular strength of this methodology is that it does not assume prior knowledge about the mechanism of toxicity and hence provides a rapid and relatively unbiased analysis of potential toxicological

■

MATERIALS AND METHODS Nanomaterials and Chemicals. Aqueous suspensions of neutral 60-nm polystyrene (PS) and 60-, 160-, and 600-nm amino-functionalized polystyrene (PS-NH2) nanoparticles were obtained from Bangs Laboratories Inc. (Fishers, IN, USA). Their primary size was determined using TEM, ζ-potential was determined from their electrophoretic mobility, and their size distribution in the experimental media was characterized by DLS. A small molecule analogue for cationic PS-NH2, polymyxin B, was from Sigma-Aldrich. See Supporting Information for detailed description of nanomaterials and chemicals used, as well as suspension and characterization of nanomaterials prior to use in biological assays. Low-Throughput Growth Inhibition Assay. Toxicity of PS and PS-NH2 NMs and polymyxin B toward E. coli BW25113(pACYC177) (KEIO collection parent strain bearing kanamycin resistant plasmid) was determined using 384-well transparent microplates. Fifty μL of NM solutions in LB supplemented with 5% FBS and 25 μg/mL kanamycin was pipetted onto microplates, inoculated with overnight-grown culture of the parent E. coli, and then incubated at 37 °C for 40 h. The optical density of the culture was measured every 30 min. Growth curves were constructed at each NM concentration and the IC5, IC50, and IC90 values with associated 95% confidence intervals were calculated. See Supporting Information for details. Characterization of Nanomaterials Interactions with Bacteria. AFM was used to image the cells in the presence of the neutral and cationic 60-nm polystyrene NMs, the ζpotential of mixtures of NMs and E. coli were determined, and the affinities of lipopolysaccharide for 60-nm PS and PS-NH2 NMs were measured. Also, ROS induction in bacterial cells as a response of exposure to 60-nm PS or PS-NH2 NMs was measured. The detailed experimental procedures are described in the Supporting Information. Bacterial Strains for Genome-Wide High-Throughput (HT) Phenotypic Screen. Growth inhibition profile of the KEIO collection of 4159 single gene knockout mutant strains of E. coli5 was used. Growth of the mutants was compared to the kanamycin-resistant parent strain of the collection, E. coli BW25113(pACYC177). The mutant strains were stored on fifteen 384-well microplates. See Supporting Information for descriptions of procedures used for storage and maintenance of E. coli strains. Genome-Wide High-Throughput (HT) Growth Inhibition Assay with E. coli Single-Gene Deletion Mutants. A HT growth inhibition test using the E. coli KEIO collection of single gene deletion mutants was performed using 60-nm PSNH2 NM and (separately) polymyxin B. Polymyxin B and PSNH2 were diluted in LB supplemented with 5% FBS and 25 μg/mL of kanamycin in 384-well microplates; an automated liquid handling system and robotics were used to dispense the media, test compounds, and to perform inoculation of bacterial strains. (See Supporting Information for details.) Each experimental condition was performed in a minimum of four replicates. The genome-wide collection of E. coli mutants was 2399

dx.doi.org/10.1021/es203087m | Environ. Sci. Technol. 2012, 46, 2398−2405


Article

Table 1. Physico−Chemical Characteristics of Polystyrene (PS) Nanomaterials Used in This Study hydrodynamic diameter (nm) in LB + 5% FBS (%Pd)d material PSNH2 PSNH2 PSNH2 PS

advertised particle sizea

hydrodynamic diameter (nm) in DI water (%Pd)b

60

56.87 ± 2.1 (8.5)

ζ-potential (mV), in DI water

density of NH2 groups (C cm−2)c

0h

43.27

0.027

220 ± 25 (14)

24 h 224 ± 30 (18) n.d.e

160

165 ± 2.6 (16.3)

55.79

0.03

356 ± 21 (23)

600

628 ± 8.1 (12.8)

52.17

n.d.

803 ± 32 (multimodal)

n.d.

60

59.2 ± 2.8 (10.5)

−54.58

82.3 ± 0.1 (12)

n.d.

Data from manufacturer. bMeasured from 100 μg/mL NMs suspension. cCalculated according to Oshima's correlation essentially as in reference 30. dFBS = fetal bovine serum. Measured from 50 μg/mL NMs suspension. en.d. = not determined. a

what had been seen in mammalian systems previously,2 neutral PS NMs over a concentration range of 1−500 μg/mL do not exhibit any antibacterial activity and that only the smallest sized cationic PS-NH2 material was toxic to bacteria in a growth inhibition assay (Figure S2). The importance of cationic charge in mediating the nanoparticles toxicity in both prokaryotes and eukaryotes has been shown for several different NM types.2,16−18 Based on these results, the 60-nm PS-NH2 NM was used as the primary material for the remainder of this study and a cationic peptide, polymyxin B, was selected as a small molecule analog/control. The concentration−effect curve of 60-nm PS-NH2 NM proved particularly steep: the difference between the concentration at which almost no effect was observed (∼ 100 μg/mL) and the IC90 (∼ 190 μg/mL) value differed by less than a factor of 2. Steep concentration−inhibition curves are often seen in HT studies on small molecules and can arise due to multisite-binding from concentration-dependent aggregation of inhibitor, or when assays are run under conditions where the enzyme concentration is high compared to the Kd (in our case, where the binding of NM to the cells is very strong).19 To gain insights into why a steep concentration−inhibition curve is observed for PS-NH2 in E. coli, we studied the physical interactions between the bacterial cells and the PS-NH2 nanomaterial using a variety of methods. Interaction of 60-nm PS-NH2 Nanomaterial with Bacterial Cells. We found that the PS-NH2 NM bound strongly to bacterial cells. The ζ-potential of the cells, which is normally negative (−44 mV), significantly increased upon the addition of 5 μg/mL of PS-NH2 particles (Figure S3A). Strong and specific attachment of multiple cationic PS-NH2 particles to E. coli cells was also observed via AFM, even though the cells were rinsed with water after exposure to the PS-NH2 NM prior to imaging (Figure S3D). No such attachment was observed when the cells were treated analogously with neutral PS NM (Figure S3B). These experiments support a model in which tight, multisite binding of the PS-NH2 NM to E. coli results in a steep concentration−inhibition curve. Detailed studies of the interaction between the PS-NH2 NM and a major component of E. coli cell surface, lipopolysaccharides (LPS) (see Supporting Information) suggest that a major mode of binding of PS-NH2 to E. coli is via interactions between the NM and LPS and that this binding is charge related (Figure S4). Sensitivity Profiles of E. coli Gene Deletion Strains for 60-nm PS-NH2 NM. To identify mutant strains that were more sensitive to the toxic NM than the E. coli parent strain (“sensitive strains initial hits”), the entire library of strains was exposed to a NM concentration equivalent to the IC5 (100 μg/

exposed to either IC5 or IC90 concentrations of the test compounds. The growth profiles of mutants after 24 h growth at 37 °C were compared with that of the parent strain, to select initial sensitive and resistant hit mutants. To confirm the initial sensitive and resistant hit mutants, the IC50 was determined for each mutant strain and compared with that of parent strain. A schematic representation of the experimental flow of the HT growth inhibition assay is presented in Figure S5 and a detailed description of the assay, its quality control, and statistical analyses, as well as of confirmatory assays are provided in the Supporting Information. Functional Classification and Characterization of Hit Mutant Strains Identified Based on HT Growth Inhibition Assay. The functions of hit genes were determined using EcoliHub (http://ecolihub.org/) and EcoCyc (http:// ecocyc.org/) bioinformatics databases. Each gene was assigned to one main biological process GO term; we also checked whether the localization GO term for each gene was either outer membrane or plasma membrane. The DAVID database (http://david.abcc.ncifcrf.gov/)14 was used to determine functional classifications of the confirmed hit genes; medium stringency was used for classification. The STRING (“Search Tool for the Retrieval of Interacting Genes/Proteins”) (http:// string-db.org/)15 tool was used to build protein−protein interactions; high confidence (0.7) was used to select the interactions. KEGG pathway identifiers (http://www.genome. jp/kegg/) were used to visualize whether confirmed hit genes occur in common metabolic pathways of E. coli. To test specific hypotheses about why certain genes might confer sensitivity or resistance to PS-NH2, the set of confirmed sensitive hit strains or a subset thereof were tested for their outer membrane integrity and electron transport activity using confirmatory assays. Detailed methods for the following series of confirmatory assays are provided in the Supporting Information: (i) outer membrane integrity and impact of added lipopolysaccharide (LPS) on toxicity of 60-nm PS-NH2 nanomaterial; (ii) ATP content and electron transport chain (ETC) activity; and (iii) sensitivity to reactive oxygen species.

■

RESULTS AND DISCUSSION

Relative Toxicity of Polystyrene Nanomaterials with Different Surface Charges in E. coli. Prior to conducting the genome-wide phenotypic toxicity screening, we first studied a series of PS nanospheres of different sizes and charges in the E. coli parent strain: neutral (no surface charge) 60-nm polystyrene nanospheres and cationic, amino-functionalized 60-, 160-, and 600-nm polystyrene nanospheres (PS-NH2) (Table 1; Figure S1). These studies revealed that, similar to 2400



Article

clustered under the GO term protein palmitoylation (a process necessary for proper formation of a number of membrane-associated proteins) with a p-value of 3.2 × 10−18. The remaining clusters corresponded to the GO terms lipopolysaccharide biosynthesis (10 genes, p-value =4.7 × 10−23), two-component regulatory systems (4 genes, p-value = 2.1 × 10−7), ATP binding (two clusters of 5 and 6 genes with p-values of 1.6 × 10−6 and 3.1 × 10−5, respectively), ubiquinone biosynthetic process (4 genes, p-value 9.9 × 10−9), iron−sulfur containing proteins (9 genes, p-value 9.7 × 10−7), and transcription regulation (5 genes, p-value 5.5 × 10−12). Of these clusters, the most significant were those that corresponded to strains in which the deleted genes encode proteins in the outer membrane and/or that are involved in lipopolysaccharide biosynthesis. We also found significant protein−protein interaction among 208 sensitive hit genes using STRING tool (Figure 1). As with DAVID tool, a

mL) of the parent strain. In a separate screen, the entire library of strains was exposed to a NM concentration equivalent to the IC90 (220 μg/mL) of the parent strain; this second screen was used to identify those strains that were more resistant to NM than the parent strain (“robust strains initial hits”). From the set of 4159 E. coli single gene deletion mutants screened, 261 “sensitive strains initial hits” and 37 “robust strains initial hits” were identified. In confirmatory detailed growth inhibition assays (see Supporting Information for more details on confirmatory assays), 208 out of 261 of the “sensitive strains initial hits” (i.e., 80%) and 23 out of 37 of the “robust strains initial hits” (i.e., 65%) were confirmed to be more sensitive or more resistant than the parent strain, respectively (Figure S8A; Table S1). A detailed description of the process by which the initial screen and confirmatory assay conditions were optimizedwhich is responsible for the low false positive rates reported hereinis provided in the Results section of the Supporting Information. The observation that the number of E. coli gene deletion mutants that are sensitive to the NM is greater than the number of resistant mutants is consistent with an earlier study on E. coli exposed to a range of chemical and environmental stressors.9 This suggests that, in bacteria, the removal of a gene product is more likely to decrease the cellular resistance to stress than to increase it. Because the PS-NH2 concentration−inhibition curve is so steep (see above and Figure S2), a minor change in concentration can result in a significant change in growth phenotype under the initial assay conditions. While the IC50 of the parent strain was 158 ± 2.5 μg/mL, the IC50 of most (∼94%) of the confirmed sensitive mutants was >80 μg/mL (i.e., less than a 2-fold reduction in IC50 compared to the parent strain). Only 13 of the confirmed sensitive strains (indicated by rectangle in Figure S8A) exhibited an IC50 lower than 80 μg/ mL. Likewise, none of the confirmed robust strains exhibited an IC50 2-fold greater than the IC50 value of the parent strain (i.e., > 320 μg/mL) and only four of the confirmed robust strains (indicated by strains within the rectangle in Figure S8B) exhibited an IC50 more than 200 μg/mL (i.e., that was more than 25% greater than the IC50 value of the parent strain). Overview of the Functions of the Genes Deleted in Differentially Sensitive Hit Strains. To classify the genes that resulted in differential sensitivity toward 60-nm PS-NH2 NM, we used functional classification of genes by DAVID gene functional classification tool,14 protein functional interactions tool STRING (“Search Tool for the Retrieval of Interacting Genes/Proteins”),15 and metabolic “Pathway Map” functionality on the Kyoto Encyclopedia of Genes and Genomes (KEGG) Web site. None of these methods revealed significant functional connections among the 23 genes that were absent in the confirmed resistant hit strains (Table S1). Thus, we were not able to identify any significant pathways where the absence of these pathways in bacterial cells results in resistance toward cationic nanomaterials. On the other hand, strong interactions were found among the 208 genes (Table S1) that, when deleted, resulted in strains confirmed to be sensitive to the 60-nm PS-NH2 nanomaterial. The gene functional classification analysis by DAVID yielded nine distinct clusters involving 84/208 of the genes deleted in confirmed sensitive hits. The two clusters that contained the highest number of genes (39 and 12 genes, respectively) were both connected with the GO term cell membrane (p-values = 4.9 × 10−46 and 8.1 × 10−14, respectively). The group of twelve cell membrane-associated genes also had ten genes that

Figure 1. Functional interactions (colored lines) between proteins encoded by E. coli genes in which loss of functionality resulted in increased sensitivity toward 60 nm PS-NH2 nanomaterial. Interactions were analyzed using STRING 8.3 tool (http://string.embl.de/). Only those proteins involved in clusters of more than three that have significant (medium stringency; 0.7) association evidence are shown. Prior evidence for functional interactions between proteins is indicated by lines, where green indicates evidence for neighborhood, red indicates evidence for gene fusion, dark blue indicates evidence for cooccurrence, gray indicates evidence for coexpression, pink indicates experimental evidence (not otherwise specified), light blue indicates evidence from databases, and orange indicates evidence from text mining. Using this method, the following clusters of genes/proteins were identified: (A) genes encoding proteins involved in LPS biosynthesis; (B) genes encoding select transmembrane proteins; (C) genes encoding proteins involved in ubiquinone biosynthesis and functionally related proteins; (D) genes encoding proteins involved in DNA repair; and (E) genes encoding proteins involved in flagellar assembly.

significant cluster of lipopolysaccharide biosynthesis proteins (Figure 1A) and a significant cluster of transmembrane proteins (Figure 1B) were identified by STRING. In addition, the STRING analysis revealed a distinct cluster of seven genes encoding for proteins involved in ubiquinone biosynthesis (Figure 1C) and revealed associations between four proteins involved in DNA repair (Figure 1D) and four proteins important for flagellar assembly (Figure 1E). Analysis of the gene products of 208 genes which corresponded to confirmed sensitive hit strains using KEGG 2401



Article

Figure 2. Sensitivity of confirmed PS-NH2 sensitive hit genes to SDS (a measure for outer membrane integrity). (A) Correlation between IC50 values for 60-nm PS-NH2 and SDS among all 208 confirmed PS-NH2 sensitive mutant strains. Strains defective in a gene encoding for a protein in LPS biosynthesis pathway or for a transmembrane protein are shown with filled symbols and the correlation coefficient for these strains is shown. (B) Enlarged view of the region indicated in (A) within the dashed rectangle with gene names. The names of the genes participating in LPS biosynthesis are shown in bold text and are underlined.

cationic PS-NH2 material was connected with GO term cell membrane. To test whether this is because the NM destabilizes the outer membrane of E. coli, we compared the sensitivities of the confirmed hit genes for the PS-NH2 nanomaterial to the sensitivity of these strains for SDS (as a proxy for outer membrane integrity) (Figure 2). Generally, the mutant strains with high sensitivity toward SDS (i.e., with low membrane integrity compared to the parent strain) were also more susceptible to PS-NH2 material, but not necessarily vice versa. This observation suggests that disruption of the integrity of the outer membrane is a critical mechanism of toxicity for PS-NH2 NM, but is not the only mechanism. The gene deletion strains that were highly sensitive to both SDS and PS-NH2 NM (relative to the parent strain) included mutants lacking components of the main transmembrane channels (tolQ, tolC, and acrB) and several mutants that were deficient in lipopolysaccharide (LPS) biosynthesis pathway (rfa mutants and gmhB) (Figure 2B). The correlation between SDS and PSNH2 sensitivity among these mutants (filled dots in Figure 2A) showed a steep slope and an r-value of 0.523 (p < 0.05). Role of Lipopolysaccharides Biosynthesis Genes in Bacterial Resistance toward 60-nm PS-NH2 Nanomaterial. The LPS biosynthesis genes which, when deleted, showed the highest sensitivity to PS-NH2 NM (rfaF, rfaC, rfaE, rfaD, rfaH, rfaG, rfaP, and ghmB) are all involved in adding the core oligosaccharides to lipid A in the LPS. Interestingly, the oligosaccharide core of LPS carries additional negative charge20 and hence may be particularly important in binding of the positively charged PS-NH2 NM. Specific binding of cationic small molecules (antimicrobial peptides) by bacterial LPS is shown here in Figure S3D and has been demonstrated in several earlier studies.21,22 Papo and Shai also demonstrated that, due to tight binding, LPS decreases the antimicrobial potency of antimicrobial peptides.21 To assess whether the effects of the cationic PS-NH2 NM on E. coli are related to the positive charge of this material, we also performed a genome-wide phenotypic toxicity screen with a well-studied cationic peptide, polymyxin B.23 LPS biosynthetic mutants that are involved in oligosaccharide core biosynthesis (Table S1) show a similar increased sensitivity (relative to the

tool showed that only 29 were found to have a known metabolic function. The two main metabolic pathways identified from this analysis were the lipopolysaccharide biosynthetic and ubiquinone biosynthetic pathways, which is consistent with the results from both the gene and protein functional clustering analyses. Taken together, the complementary analysis performed using the DAVID, STRING, and KEGG tools revealed that the largest number of genes contributing to bacterial sensitivity for PS-NH2 nanospheres were connected with formation and functioning of bacterial cell membrane, followed by defects in lipopolysaccharide and ubiquinone biosynthesis, flagellar formation, and DNA repair. Of these, the genes in the DNA repair cluster (recA, ligT, mutL, umuD, and rarA) had only a relatively minor effect on the sensitivity of E. coli toward PSNH2 (IC50 values of these mutants were 67.1−89.9% of the IC50 values of the parent strain; Table S1). All of the mutants corresponding to other biological processes (i.e., other than DNA repair) exhibited stronger impacts on cell growth in the presence of PS-NH2. These genes appear also tightly interconnected in that they encode proteins that are either located in or related to the E. coli cell wall. Thus, we hypothesized that the proper formation, stability, and functionality of the bacterial cell wall are essential for the bacteria’s ability to cope with cationic nanomaterial-induced stress. We further performed detailed confirmatory experiments to test this hypothesis (see below). In addition, because ubiquinone is involved in so many important cellular processes, we hypothesized that the sensitivity of the ubiquinone biosynthetic mutants to the cationic PS-NH2 nanomaterial could be due to any of the following reasons: decreased cellular energy (ATP) production, decreased electron transport chain activity, or increased sensitivity to ROS upon deletion of critical uniquinone biosynthetic genes. Further experiments were performed to test these hypotheses as well (see below). Importance of Cell Wall Integrity in Bacterial Resistance toward 60-nm PS-NH2 Nanomaterial. The main cluster of genes (containing 61% of the confirmed sensitive strains for which the GO localization term was known) that proved important for bacterial sensitivity toward 2402



Article

nanoparticles to membranes in vitro has been shown to induce the formation of 15−60-nm holes in lipid bilayers.25,26 Consistent with this proposed mechanism, external addition of LPS was observed to significantly decrease the toxicity of the PS-NH2 NM toward the rfaE mutant (Figure S9); we previously showed that rfaE had highest sensitivity to the PSNH2 nanomaterial among the rfa mutants. Because LPS biosynthesis is highly conserved among gram-negative bacteria, we would expect homologues to the rfa mutants to show up as more sensitive to cationic NMs in other, more environmentally relevant, gram-negative bacteria as well. Other Strains that Correlate with Decreased Membrane Stability. In addition to genes encoding transmembrane proteins and LPS biosynthetic proteins, four genes from other pathways that also showed sensitivity for SDS and PSNH2 were identified: hfq and ydiB, ygdQ and yciM. The hfq mutant showed moderate sensitivity toward SDS but very high sensitivity toward PS-NH2. This gene encodes for a protein that regulates the stability of mRNAs, especially those of stressrelated genes.27 Nichols et al.9 have shown that hfq also has a functional relationship with several genes that are important to cell membrane stability (e.g., tolC, arcA, and arcB). The deletion mutant of hfq has also been demonstrated sensitive toward a series of antibiotics and genotoxic agents.9 In this study (see below), we demonstrated that the hfq gene deletion strain is also remarkably more sensitive toward H2O2 than the parent strain (Figure 4), which is indicative of another important pathway in PS-NH2 toxicity. By contrast, the strains in which either ydiB, ygdQ, or yciM was deleted showed moderate sensitivity toward SDS but relatively low sensitivity towards the PS-NH2 NM (Figure 2B). The functions of these genes are relatively poorly defined: one of the genes, ydiB, encodes a quinate dehydrogenase which is important in oxidation reduction process, and the functions of ygdQ and yciM are not known. As mentioned above, not all the mutants with increased sensitivity toward PS-NH2 also showed an increased (compared to parent strain) SDS-sensitivity phenotype (Table S1, Figure 2). Hence, decreased outer membrane integrity is not the only

parent strain) for PS-NH2 and polymyxin B (filled circles in Figure 3; R2 = 0.79). These data suggest that LPS facilitates the

Figure 3. Sensitivity of confirmed PS-NH2 sensitive hit genes to the cationic peptide polymyxin B. Strains where genes encode proteins in the lipopolysaccharide (LPS) biosynthetic pathway (●), proteins in the ubiquinone biosynthestic pathway (▲), transmembrane proteins (◆), proteins involved in flagellar formation (■), and DNA repair (□) are deleted are indicated; all other strains are indicated with an unfilled diamond (◇). The trendline and correlation coefficient (r = 0.89) between IC50 values for PS-NH2 and polymyxin B is shown for LPS biosynthesis genes.

binding of cationic NMs, similar to what others have observed for cationic peptides.24 We hypothesize that, by binding to cationic substances, the LPS prevents direct contact between the cationic toxins with the bacterial outer membrane, and hence provides a level of protection for the cell. Given that the effects of cationic polystyrene nanomaterial and cationic polymyxin B were similar in our assay, we predict that other classes of positively charged NMs of similar size will likely elicit a similar response. Indeed, direct exposure of other polycationic

Figure 4. Sensitivity of confirmed PS-NH2-sensitive strains to ROS (H2O2). (A) Correlation between IC50 values for PS-NH2 and H2O2 among all 208 strains confirmed to be sensitive to PS-NH2; (B) enlarged view of the region indicated with dashed rectangle in (A) with gene names. The genes participating in ubiquinone biosynthesis are shown with filled symbols. 2403



Article

activity. Because the ubiquinone pathway is highly conserved, we predict that similar effects should be seen in a wide range of other organisms, including other, more environmentally relevant systems and also eukaryotes. Taken together, these data suggest that in addition of the disruption of the outer membrane integrity, production of ROS is an important pathway of toxicity for PS-NH2. Moreover, our finding that there is at least one mutant for which there is an additive effect between ROS sensitivity and disruption of membrane integrity (hfq), shows that these two toxicity mechanisms do not appear redundant (i.e., disruption of membrane integrity is not necessarily just due to generation of ROS by the NM). ROS generation has been demonstrated as one of the main toxicity mechanisms of cationic polystyrene nanomaterials also in mammalian cells.2,29

physiological function that results in bacterial sensitivity toward cationic nanomaterials. A detailed study of other potential physiological functions with the special emphasis on cellular redox reactions was conducted and is discussed below. Importance of Membrane Redox Reactions and Ubiquinone Biosynthesis Pathway in Bacterial Resistance toward 60-nm PS-NH2 Nanomaterial. Strains defective in genes involved in ubiquinone biosynthesis were studied in more detail because this group of genes was identified both as a significant functional cluster in the protein− protein functional analysis as well as in the metabolic pathway analysis (Figure 1). Also, three of the mutants deficient in ubiquinone biosynthesis (ubiE, ubiG, and ubiF) were among the twenty most sensitive confirmed hit genes (Table S1). Ubiquinone plays a central role in cellular electron transport chain (ETC) activity being one of the electron carriers, and is also known as a powerful antioxidant.28 Therefore, we developed three possible hypotheses to explain the increased sensitivity of ubiquinone biosynthetic mutants to PS-NH2: that deletion of genes involved in ubiqinone biosynthesis results in increased sensitivity to PS-NH2 (1) due to decreased cellular energetic level, (2) due to decreased electron transport chain activity, and/or (3) due to increased susceptibility to oxidative stress. The first hypothesis was tested by determining the ATP levels of confirmed PS-NH2 sensitive hit strains. However, no significant correlation between sensitivity toward PS-NH2 and the cellular ATP level was observed (Figure S10). To test the second hypothesis, we determined whether the sensitivity to PS-NH2 of the ubiquinone biosynthetic mutants is related to their modified ETC activity (Figure S11A). Indeed, addition of the PS-NH2 nanomaterial to the parent E. coli strain induced an increase in bacterial ETC activity (Figure S11B). Thus, we suggest that the increased flow of electrons in the presence of the PS-NH2 NM cannot be handled in ubiquinone-deficient mutants, resulting in their increased sensitivity to PS-NH2. This conclusion is supported by the fact that not only ubi mutants but also visC and nuoJ gene deletion mutants (which encode for oxidoreductases that are important for electron transfer to ubiquinone) exhibited lower electron transport chain activity and also sensitivity towards the PS-NH2 NM (Table S1). To test the third hypothesis, that the ubiquinone gene deletion strains are more sensitive to the PS-NH2 NM due to increased oxidative stress, we analyzed the amount of intracellular ROS formed in response to bacterial exposure to the PS-NH2 NM and the quenching of this ROS by an antioxidant (Nacetylcysteine). Indeed, higher levels of PS-NH2 were required to increase ROS levels in E. coli parent strain than in the ubi, visC, and nuoJ mutants (Figure S12A). Addition of Nacetylcysteine decreased the toxicity of the PS-NH2 material to both the parent E. coli strain and to the ubiF mutant strain (selected as a representative for ubi mutants) (Figure S12B and S12C). These observations suggest that the addition of the PSNH2 material to E. coli cells induces oxidative stress and that the presence of intact ubiquinone can mitigate this stress. Indeed, the ubiquinone biosynthetic genes mutants that are sensitive to PS-NH2 NM were also shown to be more sensitive toward ROS (shown as sensitivity to H2O2 in Figure 4). Thus, the results reported herein are consistent with a mechanism in which addition of the PS-NH2 nanomaterial to E. coli cells induces oxidative stress and ubiquinone reduces this stress either through direct interaction between ubiquinone with ROS and/or due to the role that ubiquinone plays in normal ETC

■

ASSOCIATED CONTENT

* Supporting Information S

Detailed Materials and Methods, a Results section describing the optimization and performance of the HT bacterial growth assay, one table and 12 figures illustrating the optimized HTS workflow, characterization of hit mutants with differential sensitivites to the 60-nm PS-NH2 nanomaterial, and confirmation of toxicity mechanisms of this nanomaterial. This information is available free of charge via the Internet at http:// pubs.acs.org/.

■ ■

AUTHOR INFORMATION

Corresponding Author

*Phone: (310) 794-9112; e-mail: [email protected].

ACKNOWLEDGMENTS This material is based upon work supported by the University of California Center for Environmental Implications of Nanotechnology in a grant from the National Science Foundation and the Environmental Protection Agency under Cooperative Agreement DBI-0830117. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation or the Environmental Protection Agency. This work has not been subjected to EPA review and no official endorsement should be inferred. A.I. received financial support from European Science Foundation program Mobilitas grant MJD67. We acknowledge the use of the SPM facility at the Nano and Pico Characterization Lab at the California NanoSystems Institute.

■

REFERENCES

(1) Behra, R.; Krug, H. Nanoecotoxicology: Nanoparticles at Large. Nat. Nanotechnol. 2008, 3 (5), 253−254. (2) Xia, T.; Kovochich, M.; Liong, M.; Zink, J. I.; Nel, A. E. Cationic Polystyrene Nanosphere Toxicity Depends on Cell-Specific Endocytic and Mitochondrial Injury Pathways. ACS Nano 2007, 2 (1), 85−96. (3) Bhattacharjee, S.; de Haan, L. H.; Evers, N. M.; Jiang, X.; Marcelis, A. T.; Zuilhof, H.; Rietjens, I. M.; Alink, G. M. Role of surface charge and oxidative stress in cytotoxicity of organic monolayer-coated silicon nanoparticles towards macrophage NR8383 cells. Part. Fibre Toxicol. 2010, 7 (25). (4) Lyon, D. Y.; Alvarez, P. J. J. Fullerene Water Suspension (nC60) Exerts Antibacterial Effects via ROS-Independent Protein Oxidation. Environ. Sci. Technol. 2008, 42 (21), 8127−8132. (5) Baba, T.; Ara, T.; Hasegawa, M.; Takai, Y.; Okumura, Y.; Baba, M.; Datsenko, K. A.; Tomita, M.; Wanner, B. L.; Mori, H. Construction of Escherichia coli K-12 in-Frame, Single-Gene Knockout

2404



Article

Mutants: The Keio Collection. Mol. Syst. Biol. 2006, Article number: 2006.0008. (6) Becket, E.; Chen, F.; Tamae, C.; Miller, J. H. Determination of Hypersensitivity to Genotoxic Agents Among Escherichia coli Single Gene Knockout Mutants. DNA Repair 2010, 9 (9), 949−957. (7) Desai, K. K.; Miller, B. G. Recruitment of Genes and Enzymes Conferring Resistance to the Nonnatural Toxin Bromoacetate. Proc. Natl. Acad. Sci., U.S.A. 2010, 107 (42), 17968−17973. (8) Liu, A.; Tran, L.; Becket, E.; Lee, K.; Chinn, L.; Park, E.; Tran, K.; Miller, J. H. Antibiotic Sensitivity Profiles Determined with an Escherichia coli Gene Knockout collection: Generating an Antibiotic Barcode. Antimicrob. Agents Chemother. 2010, AAC.00906−00909. (9) Nichols, R. J.; Sen, S.; Choo, Y. J.; Beltrao, P.; Zietek, M.; Chaba, R.; Lee, S.; Kazmierczak, K. M.; Lee, K. J.; Wong, A.; Shales, M.; Lovett, S.; Winkler, M. E.; Krogan, N. J.; Typas, A.; Gross, C. A. Phenotypic Landscape of a Bacterial Cell. Cell 2011, 144 (1), 143− 156. (10) Sharma, O.; Datsenko, K. A.; Ess, S. C.; Zhalnina, M. V.; Wanner, B. L.; Cramer, W. A. Genome-wide Screens: Novel Mechanisms in Colicin Import and Cytotoxicity. Mol. Microbiol. 2009, 73 (4), 571−585. (11) Tamae, C.; Liu, A.; Kim, K.; Sitz, D.; Hong, J.; Becket, E.; Bui, A.; Solaimani, P.; Tran, K. P.; Yang, H.; Miller, J. H. Determination of Antibiotic Hypersensitivity among 4,000 Single-Gene-Knockout Mutants of Escherichia coli. J. Bacteriol. 2008, 190 (17), 5981−5988. (12) Ji, Z.; Jin, X.; George, S.; Xia, T.; Meng, H.; Wang, X.; Suarez, E.; Zhang, H.; Hoek, E. M. V.; Godwin, H.; Nel, A. E.; Zink, J. I. Dispersion and Stability Optimization of TiO2 Nanoparticles in Cell Culture Media. Environ. Sci. Technol. 2010, 44 (19), 7309−7314. (13) Gou, N.; Gu, A. Z. A New Transcriptional Effect Level Index (TELI) for Toxicogenomics-based Toxicity Assessment. Environ. Sci. Technol. 2011, 45 (12), 5410−5417. (14) Dennis, G.; Sherman, B. T.; Hosack, D. A.; Yang, J.; Gao, W.; Clifford Lane, H.; Lempicki, R. A. DAVID: Database for Annotation, Visualization, and Integrated Discovery. Genome Biol. 2003, 4, P3. (15) Snel, B.; Lehmann, G.; Bork, P.; Huynen, M. A. STRING: a Web-Server to Retrieve and Display the Repeatedly Occurring Neighbourhood of a Gene. Nucleic Acids Res. 2000, 28 (18), 3442− 3444. (16) El Badawy, A. M.; Silva, R. G.; Morris, B.; Scheckel, K. G.; Suidan, M. T.; Tolaymat, T. M. Surface Charge-Dependent Toxicity of Silver Nanoparticles. Environ. Sci. Technol. 2010, 45 (1), 283−287. (17) Goodman, C. M.; McCusker, C. D.; Yilmaz, T.; Rotello, V. M. Toxicity of Gold Nanoparticles Functionalized with Cationic and Anionic Side Chains. Bioconjugate Chem. 2004, 15 (4), 897−900. (18) Kim, J. S.; Kuk, E.; Yu, K. N.; Kim, J.-H.; Park, S. J.; Lee, H. J.; Kim, S. H.; Park, Y. K.; Park, Y. H.; Hwang, C.-Y.; Kim, Y.-K.; Lee, Y.S.; Jeong, D. H.; Cho, M.-H. Antimicrobial Effects of Silver Nanoparticles. Nanomedicine: NBM 2007, 3 (1), 95−101. (19) Shoichet, B. K. Interpreting Steep Dose-Response Curves in Early Inhibitor Discovery. J. Med. Chem. 2006, 49 (25), 7274−7277. (20) Müller-Loennies, S.; Lindner, B.; Brade, H. Structural Analysis of Oligosaccharides from Lipopolysaccharide (LPS) of Escherichia coli K12 Strain W3100 Reveals a Link between Inner and Outer Core LPS Biosynthesis. J. Biol. Chem. 2003, 278 (36), 34090−34101. (21) Papo, N.; Shai, Y. A Molecular Mechanism for Lipopolysaccharide Protection of Gram-negative Bacteria from Antimicrobial Peptides. J. Biol. Chem. 2005, 280, 10378−10387. (22) Rosenfeld, Y.; Shai, Y. Lipopolysaccharide (Endotoxin)-host Defense Antibacterial Peptides Interactions: Role in Bacterial Resistance and Prevention of Sepsis. BBA - Biomembranes 2006, 1758 (9), 1513−1522. (23) Storm, D. R.; Rosenthal, K. S.; Swanson, P. E. Polymyxin and Related Peptide Antibiotics. Annu. Rev. Biochem. 1977, 46, 723−763. (24) Liu, L.; Xu, K.; Wang, H.; Tan, J. P. K; Fan, W.; Venkatamaran, S. S.; Li, L.; Yang, Y.-Y. Self-Assembled Cationic Peptide Nanoparticles as an Efficient Antimicrobial Agent. Nat. Nanotechnol. 2009, 4, 457− 463.

(25) Leroueil, P. R.; Berry, S. A.; Dthie, K.; Han, G.; Rotello, V. M.; McNerny, D. Q.; Baker, J. R.; Orr, B. G.; Banaszak Holl, M. M. Wide Varieties of Cationic Nanoparticles Induce Defects in Supported Lipid Bilayers. Nano Lett. 2008, 8, 420−424. (26) Verma, A.; Stellacci, F. Effect of Surface Properties on Nanoparticle-cell Interactions. Small 2010, 1, 12−21. (27) Gottesman, S. The small RNA Regulators of Escherichia coli: Roles and Mechanisms. Annu. Rev. Microbiol. 2004, 58, 303−328. (28) Kawamukai, M. Biosynthesis, Bioproduction and Novel Roles of Ubiquinone. J. Biosci. Bioeng. 2002, 94 (6), 511−517. (29) Xia, T.; Kovochich, M.; Brant, J.; Hotze, M.; Sempf, J.; Oberley, T.; Sioutas, C.; Yeh, J. I.; Wiesner, M. R.; Nel, A. E. Comparison of the Abilities of Ambient and Manufactured Nanoparticles To Induce Cellular Toxicity According to an Oxidative Stress Paradigm. Nano Lett. 2006, 6 (8), 1794−1807. (30) Ohshima, H. Approximate Analytic Expression for the Electrophoretic Mobility of a Spherical Colloidal Particle. J. Colloid Interface Sci. 2001, 239, 587−590.

2405


Genome-Wide Bacterial Toxicity Screening ... - ACS Publications

Recommend Documents