Transcriptome profile with 20 nm silver nanoparticles in yeast

Abstract

Engineered nanomaterials are commercially used in everyday products including zinc sunscreens and water-resistant fabrics and surfaces. Therefore, understanding the effects of engineered nanomaterials on the environment is crucial for the responsible use of these technologies. We investigated the effects of 20 nm spherical citrate-coated silver nanoparticles (AgNPs) on the budding yeast Saccharomyces cerevisiae. Our growth assay showed that AgNPs have an inhibitory effect on yeast growth with concentrations above 5 μg/mL. Hundreds of genes in AgNP-treated cells were differentially expressed according to our transcriptome analysis based on RNAseq, including genes implicated in rRNA processing, ribosome biogenesis, cell wall formation, cell membrane integrity and mitochondrial functions. In particular, genes whose functions are associated with processing of small and large subunits of ribosomes were upregulated, while genes for cell wall/plasma membrane/mitochondrial integrity were downregulated. Consistently, our cell wall stability assay confirmed that cells with AgNPs are more susceptible to cell wall damage than non-treated cells. Levels of four significantly altered genes with AgNPs, including FAF1, SDA1, TIR1 and DAN1, were validated by reproducible results with RT-qPCR assays. Our transcriptome profile leads us to conclude that the exposure of cells to sublethal amounts of AgNPs affects many cellular processes negatively.

ENMs, yeast, toxicology, gene expression

INTRODUCTION

Engineered nanomaterials (ENMs) are valuable and unique due to their small size, large surface-area-to-volume ratio, aggregation, chemical composition, solubility and shape (Nel et al.2006). They are currently used in over 1000 commercially available products, ranging from sunscreens to water-resistant surfaces (Damoiseaux et al.2011). ENMs have been predicted to play a part in future-targeted disease treatment, nanorobotics and next-generation electronics. Although the application of ENMs is increasing almost daily, our understanding of their toxicity is lagging behind the technology, thus encumbering safe and rapid deployment of these materials. In order to responsibly use these technologies, understanding how ENMs interact with the environment is a necessity. Their size allows them to interact at a molecular level, making them potentially useful in many applications; however, it also makes them potentially dangerous.

Nearly every class of ENMs has been found to have some negative biological effects; the question however is whether or not these effects pose a significant health or environmental risk (Colvin 2003; Kwolek-Mirek and Zadrag-Tecza 2014). In order to determine whether a particular ENM is toxic or not, one must consider the specific properties of the materials being investigated, such as size, shape, etc. Studies have shown that even between batches of nanomaterials, toxicity may vary. Many studies in academia and industry have attempted to develop methods of understanding the toxicity of ENMs. Gold nanoparticles (GNPs) have gained considerable attention for potential application in cancer treatment such as photothermal therapy (Riley and Day 2017), and it was found that GNPs of different sizes are not inherently toxic to human cells including keratinocytes and leukemia (Connor et al.2005; Patra et al.2007). However, 2 nm GNPs functionalized with both cationic and anionic surface groups were toxic (Goodman et al.2004). Another metal nanomaterial that has been widely used in a range of biomedical applications, including diagnosis, treatment, drug delivery and medical device coating, is silver nanoparticles (AgNPs) (Ge et al.2014). While it is well known that AgNPs exhibit antibacterial (Kim et al.2007), antifungal (Kim et al.2008), antiviral (Sun et al.2005) and anti-inflammatory properties (Nadworny et al.2008), there exists a report demonstrating that AgNPs at 100 μg/mL with different sizes in HaCat cells are not toxic (Ray, Yu and Fu 2009). Although the mechanism underlying their toxicity in different cells and organisms is not yet fully understood, the general consensus is that AgNPs cause cell membrane disruption (Sondi and Salopek-Sondi 2004; Kim et al.2009; Madhavan et al.2014), as well as oxidative stress (Garcia-Saucedo et al.2011; Nogueira et al.2014).

An emerging ideal tool to assess how an organism responds to a spectrum of ENMs is RNAseq analysis that involves quantification of the expression of its genome. This high-throughput genomic/transcriptomic technology has been implemented for investigating the effects of AgNPs in aquatic organisms, soil invertebrates, green algae and bacteria (Simon et al.2013; Boenigk et al.2014; Novo et al.2015; Orsini et al.2017; Sun et al.2017) and revealed that AgNPs cause differential expression of transcripts encoding components of the cell wall. Consistently, recent cell biological analyses with Candida albicans, a pathogenic fungus, have shown that AgNPs disrupt the membrane (Kim et al.2009), induce apoptosis (Hwang et al.2012) and cause ultrastructural changes (Vazquez-Munoz, Avalos-Borja and Castro-Longoria 2014). However, little is known about transcriptomic profiles of fungal cells treated with AgNPs. Therefore, the present study has used the budding yeast, Saccharomyces cerevisiae, to assess effects of AgNPs on the transcriptional activities of individual genes, offering a comprehensive picture of cellular function in the presence of AgNPs. The rationale for the use of S. cerevisiae was that it is one of the simplest eukaryotic organisms, but carries genes and their corresponding proteins that function in a spectrum of biological processes taking place in our cells. Another purpose of this study is to provide a standard operating procedure for assessing the effects of nanomaterials on fungal organisms before they are put into the environment. We reveal here that a sublethal amount (5 μg/mL) of 20 nm spherical AgNPs in yeast culture leads to a significant change in transcriptome profile when compared with non-treated cell culture, supporting the notion that AgNPs are environmental stress factors.

MATERIALS AND METHODS

Ag nanoparticles

PELCO^® NanoXact AgNPs (20 nm) suspended in 2 mM sodium citrate solution (pH 7.6) at a concentration of 20 μg/mL were obtained from Ted Pella, Inc., (Redding, CA). The average diameter of the spheroidal nanoparticles (NPs) is 20 ± 2.9 nm, measured by a JOEL 1010 Transmission Electron Microscope (for more chemical and physical information, visit www.nanocomposix.com). This sample of NPs shows the absorption band at 393 nm (www.nanocomposix.com).

Growth assay with exposure to AgNPs

Wild-type S. cerevisiae S288C cells (MATα SUC2 gal2 mal2 mel flo1 flo8–1 hap1) were purchased from ATCC (American Type Culture Collection) (Manassas, VA) and were grown in synthetic-defined glucose (SD-Glu) media overnight in a shaker incubator at 30°C. The optical density (OD) at 600 nm of the cultures was measured using a BioMate 3S spectrophotometer (Thermo Fisher Scientific, Waltham, MA) after being cultured for 16–18 h in the shaker incubator. Cultures were grown to 1 × 10⁷ cells/mL and then inoculated into 2X SD-Glu to an OD of 0.006. These cells were then plated on a 96-well culture plate, along with AgNPs at concentrations of 0.05, 0.1, 0.5, 1, 2, 5 and 10 μg/mL. Positive controls containing no nanomaterials as well as blank wells containing no cells were also created. Each experimental trial, positive control and blank were replicated four times on each plate. Cells were then placed on an ELx808 Absorbance Microplate Reader (BioTek, Winooski, VT), and grown for 30 h at 30°C, with the OD at 600 nm being recorded every 10 min. OD readings from the blank wells were subtracted from each test condition OD, and the resulting ODs were averaged to create representative growth curves for each test concentration. The logarithmic portion of the growth curves was used to determine doubling time for each NP concentration. This growth assay was repeated three times.

Metabolic activity assessment using FUN-1 dye

FUN-1 cell stain dye, a viability probe for fungal cells, was purchased from Thermo Fisher Scientific. Yeast cells (S288C) were grown for 16–18 h to an OD of 1.0 in SD-Glu media. The following day second inoculation was made and once again grown for 16–18 h to an OD of 1.0. The cells were then inoculated into SD-Glu media to an OD of 0.2 with an AgNP concentration of 0, 5 or 10 μg/mL and were grown for 5 h at 30°C in a shaker incubator. The cells were spun and transferred to 0.2 μm filtered water containing 2% D-glucose and 10 mM Na-HEPES and stained using 1 μL of 10 mM FUN-1 stock solution (final concentration of 20 μM) for 30 min. Stained cells were then examined using an Olympus IX81 inverted fluorescent microscope equipped with an ORCA camera (Hamamatsu, Bridgewater, NJ) with the excitation/emission filter set at 480/620 nm.

Total RNA extraction

Yeast cells (S288C) were grown in SD-Glu media to mid-log phase corresponding to an OD at 600 nm of 0.3–0.6. These cells were incubated at 30°C with shaking at 220 rpm with either SD-Glu media only for the control or SD-Glu media containing 5 μg/mL of AgNPs for 5 h. These experiments were performed in triplicates. Total RNA was extracted with the protocol and materials from RiboPure Yeast RNA Extraction Kit (Thermo Fisher Scientific) on three control and three AgNP-treated samples. The RNA concentration was calculated by measuring the OD at 280 nm using NanoPhotometer^® P330 (v1.0, Implen, Westlake Village, CA) or the Qubit 3.0 Fluorometer. Final concentrations of total RNA ranged from 960 to 1200 ng/μL.

mRNA isolation and cDNA synthesis

mRNA was isolated from the total RNA, using TruSeq^® Stranded mRNA LT Sample Preparation Kit (Illumina, San Diego, CA) by following the Low Sample Protocol. The first strands of cDNA were synthesized from the purified mRNA, using SuperScript II Reverse Transcriptase from the kit, followed by synthesis of the second strand of cDNA. Each cDNA sample was ligated with a distinct adaptor for sequencing, and the ligated cDNA fragments on both ends were amplified for 15 cycles. The end products were suspended in 30 μL Resuspension Buffer with final concentrations ranging from 45 to 60 ng/μL. The enriched cDNA libraries were sequenced using an Illumina hiseq 2500 Sequencing system (Kansas Medical Genome Center). One hundred nucleotides from only one end of each sequence (single-end sequencing) were completed with the cDNA libraries originated from three control and three AgNP-treated cells.

Analysis of sequencing data

Data from cDNA sequencing were analyzed using Galaxy, a web-based platform for sequencing analysis (https://Usegalaxy.org). The files received from Kansas University Medical Center were uploaded to the Galaxy server and were concatenated, resulting in a single file for each sample containing all sequencing data related to that sample. A quality check was run on each file to ensure high-quality reads. The files were then groomed to convert them into Sanger format, which is needed for the following steps. To ensure high fidelity, the reads were trimmed based on quality, and any bases with a quality score below 20 were removed from the reads. To eliminate bias of primers and to ensure removal of adapters, 12 bases were trimmed from the 5^΄ end of the reads. The reads were then filtered, resulting in any reads shorter than 80 base pairs being removed. Reads were then aligned to the S. cerevisiae reference genome (S288C) downloaded from the Saccharomyces Genome Database (SGD) using Tophat in Galaxy. The transcriptome was assembled using Cufflink using a reference annotation matching the reference genome downloaded from SGD. Finally, using Cuffdiff the aligned sequences expression rates were compared between conditions, resulting in a list of differentially expressed genes. Genes with a P-value (q) ≤ 0.05 were considered to be differentially expressed. The differentially expressed genes were then grouped into categories based on their correlating Gene Ontology terms (GO terms) according to the SGD (YeastGenome.org).

Quantitative reverse transcription PCR

Total RNA samples isolated from three control and three experimental (AgNP-treated) cell cultures were used to produce cDNA with the Verso cDNA conversion kit (Thermo Fisher Scientific). The resulting cDNA concentration was quantified with the Qubit 3.0 Fluorometer. A primer efficiency test was performed to validate DNA primers and cDNA samples for reverse transcription quantitative PCR (RT-qPCR) experiments. FAF1, SDA1, DAN1, TIR1 and ALG9 primers were chosen for this test. FAF1, SDA1, DAN1 and TIR1 genes were chosen because they were differentially expressed in AgNP-treated samples based on our RNAseq data, while the expression of ALG9, a housekeeping gene, was not affected by the presence of the NP. In this test, serially diluted cDNA (dilution factor of 5) samples with or without a fixed amount of primers were subjected to PCR amplification using GoTaq qPCR kit (Promega, Madison, WI). The primer efficiency and R-squared values were calculated with MxPro^® software (Agilent, Santa Clara, CA). Our R-squared values for the five primer sets were between 0.99 and 1.00, indicating good precision in the preparation of the dilution assay. Primer efficiency values for the five primers were ranging from 1.69 to 1.74. After this efficiency test, 60 ng of cDNA from three control and three experimental samples was used as a template for amplification of the cDNAs for the five genes (FAF1, SDA1, DAN1, TIR1 and ALG9) by following the GoTaq qPCR Master Mix protocol (Promega). For each target gene to be amplified, we prepared a non-treated control reaction mixture that contains all the same reagents including primers, GoTaq Master mix and water but lacking any cDNA. After each well contained all reagents, primers, cDNA and water, they were thoroughly mixed by pipetting up and down with a new sterilized pipette tip for each individual well. After being mixed, the plate was capped and centrifuged for 1 min. Then, the centrifuged plate was placed in the pre-heated MX3005p machine for PCR amplification. The pfaffl method was utilized to determine the relative fold change in gene expression of a target gene in comparison to the ALG9 gene (Pfaffl 2001). The relative expression ratio of the target gene is calculated based on E (RT-PCR efficiencies) and CP (crossing point) deviation versus the control. Then the gene's expression was compared to that of ALG9 gene.

Measurement of reactive oxygen species

Reactive oxygen species (ROS) and superoxide levels were quantified with flow cytometry, and the experiment was performed twice in triplicate with each experiment testing 18 samples. Each cell sample was diluted to have an OD of 0.1 cells in 100 μL of SD-Glu media and incubated in a shaking incubator for 6 h at 30°C. At hour 6, 5 μg/mL of dihydrorhodamine 123 (DHR123) or dihydroethidium (DHE) was added, and the samples were incubated for another 2 h before taken to 1 mL with 1XPBS buffer and quantified with flow cytometry (Attune NxT acoustic focusing cytometer, Life Technologies). Samples 1–9 tested for the presence of ROS by utilizing the ROS indicator DHR123. Samples 10–18 tested for the presence of superoxide similarly by utilizing the superoxide indicator DHE. Samples 1–3 and 10–12 had a concentration of 0 μg/mL of AgNPs, samples 4–6 and 13–15 had a concentration of 5 μg/mL AgNPs, and samples 7–9 and 16–18 had a concentration of 10 μg/mL AgNPs. Once recorded with the cytometer, the data were gated. The gates were customized to account for the % fluorescence of each target indicator (DHR123 and DHE) in their respective sample.

Cell wall stability assay

The cell wall integrity was tested with a cell-wall-degrading enzyme, Zymolase 100T, in triplicate. Each cell sample was diluted to an OD of 1.0 in a total volume of 200 μL in a 96-well plate. Each well also contained 0 or 10 μg/mL of Zymolase 100T and varying concentrations of AgNPs (0, 5 and 10 μg/mL). Once each sample was plated, the 96-well plate was inserted into the ELx808 plate reader (BioTek) and the OD at 595 nm was recorded every 10 min for 6 h.

Statistical analysis

In the cell viability assay (Fig. 2), 100 cells from non-treated, treated with 5 μg/mL and treated with 10 μg/mL samples were randomly selected, and levels of viability were determined. This viability assay was repeated in triplicate, and the mean number of viable cells in each of the three samples in each group was used to determine the final values depicted in Fig. 2. A two-tailed equal variance Student's t-test was performed, and no statistical difference was seen between the three sample groups suggesting AgNPs up to 10 μg/mL have no effect on cell viability measured by Fun1 fluorescent dye.

RT-PCR was performed three times in triplicate for two target upregulated and downregulated genes found to be statistically significant from a list of differentially expressed genes created from our RNA-seq data. The resulting Ct values from RT-PCR were incorporated in the Pfaffl method to determine each gene's differential fold change. A similar Student’s t-test was performed to determine P values.

ROS and superoxide levels were quantified with flow cytometry, and the experiment was performed twice in triplicate with each experiment testing 18 samples. The mean value for each group (0, 5 and 10 μg/mL) and its corresponding standard deviation value were determined prior to Student’s t-test to calculate P values.

RESULTS

AgNPs negatively affect yeast growth

To investigate the effects of spherical AgNPs (∼20 nm in diameter) on the yeast growth, unicellular budding yeast cells were incubated for 24 h at 30°C at varying concentrations of AgNPs (0–10 μg/mL). We observed that the mean growth rate for three non-treated controls was similar to those of AgNP-treated compared groups when the tested AgNP concentrations were below 5 μg/mL (Fig. 1A). However, a treatment with more than or equal to 5 μg/mL of AgNPs led to a significant growth rate reduction compared with non-treated controls. The mean maximum ODs at 600 nm reached by cells exposed to 10 and 5 μg/mL of AgNPs were 0.57 ± 0.2 and 1.61 ± 0.03, respectively, compared to the average maximum OD at 600 nm (1.73 ± 0.01) reached by non-treated control experiments (Fig. 1B). The average amount of time during the growth cycle spent in lag phase for each concentration below 5 μg/mL was very similar to the time spent in lag phase for the controls (10.6 ± 0.3 h) (Fig. 1C). However, the average times spent in lag phase by cells grown in 5 μg/mL (13.2 ± 0.80 h) and 10 μg/mL (19.9 ± 0.54 h) of AgNPs were significantly longer than that of the control (10.6 ± 0.3 h) (Fig. 1C). The mean doubling times of control cells (1.28 ± 0.13 h) was the shortest among all experiments, and the corresponding doubling time for each AgNP-treated culture increased in a dose-dependent manner. In particular, the doubling times of the cells exposed to AgNP concentrations of 5 and 10 μg/mL were 1.8 ± 0.24 and 2.7 ± 0.15 h, respectively (Fig. 1D). Taken together, these results suggest that exposure to AgNPs at concentrations higher than 5 μg/mL inhibits the efficient growth of exposed yeast cells. Ivask et al (2014) previously demonstrated that AgNO₃ is more toxic than AgNPs when they are similar in size (Ivask et al.2014). Therefore, we tested the effect of AgNO₃ on yeast cell growth and found that the cell culture with 0.05 μg/mL of AgNO₃ showed a slower rate of yeast growth than the non-treated control culture (Fig. S1, Supporting Information). The culture with 0.1 μg/mL did not support yeast growth, suggesting that 0.1 μg/mLAgNO₃ is minimum inhibitory concentration. Our experiment with AgNO₃ indicates that free silver ions affect more negatively on yeast proliferation.

Figure 1.

Effects of silver nanoparticles on the viability of yeast starting at a cell concentration of 1.5 × 10⁶ cells/mL. (A) Growth curves of wild-type (S288C) cells grown in SD-Glu media containing different concentrations (0–10 μg/mL) of silver nanoparticles. The graph was produced by measuring the optical density (OD) at 600 nm of the solution once every 10 min for 24 h. Each plate contained four replicate experiments. Two plates were tested. Each point on the curve is an average of eight experiments. (B) The maximum OD at 600 nm of each test concentration of silver nanoparticles, with the background subtracted. Control wells containing only media and silver are subtracted from each test well resulting in an optical density reading indicative of cell concentration. (C) Average amount of time spent in lag phase of cell cycle for each sample. (D) Doubling time of each concentration of silver nanoparticles was found using the natural log of growth curves from (A) to determine the growth rate, which was then used to calculate the doubling time. Each bar represents an average doubling time of eight experiments.

Measurement of FUN1 dye transport to the vacuole

FUN-1 dye staining assay is often used as a live/dead assay for yeast. The dye enters the cytoplasm where it emits green fluorescence. As transported to the vacuolar lumen, it self-assembles into fluorescent red cylindrical intravascuolar structures (CIVS) (Woodman et al.2016). In an attempt to understand what may cause growth defects in the presence of AgNPs, yeast cells were incubated with FUN-1 dye (20 μM, final concentration) with or without AgNPs. Yeast cells emitting green fluorescence in the cytoplasm with visible CIVS in the vacuole are considered active cells due to the process in which they transport the dye to the vacuole (Fig. 2A, left), whereas cells emitting green fluorescence in the cytoplasm with no visible CIVS are considered metabolically inactive or presumed to be dead (Fig. 2A, right). From all yeast cells exposed to 0, 5 and 10 μg/mL of AgNPs for 5 h, we found no significant difference in the percentage of cells showing the metabolic activity of FUN1 dye transfer to form CIVS between the exposure concentrations and the control (Fig. 2B and C). Regardless of the presence or absence of AgNPs, at any concentration, over 90% of the cells were found to have no defect in transporting FUN-1 dye to the vacuole, indicating no effect of AgNP treatment on the trafficking of FUN-1 dye to the vacuole.

Figure 2.

FUN-1 stain to determine metabolic activity of silver-nanoparticle-treated yeast cells. (A) Example of a metabolically active yeast cell stained with FUN-1 (left). Example of a metabolically inactive and presumably dead yeast cells treated with FUN-1 dye (right). (B) Control cells with no AgNP treatment, cells treated with 5 μg/mL AgNPs and cells treated with 10 μg/mL AgNPs were incubated for 30 min with FUN-1 dye prior to visualization on the fluorescence microscope. Images were taken under GFP and Texas Red channels, and merged to form the composite image shown. (C) Quantification of percentage of cells considered metabolically active based on their successful transport of FUN-1 dye to the vacuole.

cDNA sequencing reveals up- and downregulated genes with AgNPs

In lieu of harnessing 1000 different metabolic assays to identify changes occurring in the cell treated with AgNPs, we wished to determine differential gene expression that may have occurred in response to 5 μg/mL of AgNP exposure by examining the transcriptional profiles of S. cerevisiae. We utilized RNA-Seq (sequencing of cDNA) to determine the expression profiles of both our control and AgNP-treated cells. Cells were grown and tested in triplicates, and we then performed a total RNA extraction, isolated mRNA and converted it into cDNA for sequencing (Fig. 3A). After trimming and processing through Galaxy, a total of 42 808 285 accepted reads were obtained from the three control and three AgNP experiments. Of these clean reads, an average of 92.0 and 93.5% of the total reads mapped to the reference genome (S288C) in the absence and presence of AgNPs, respectively, indicating that the sequenced reads accurately reflect the transcriptional expression of S. cerevisiae. We identified 7126 genes (including non-coding cDNA), of which expression levels of 1845 genes in AgNP-treated samples were found to be statistically different from those of non-treated control samples (q-values < 0.05) (Table 1). Of these, 1077 genes were upregulated and 768 genes were downregulated. Our GO analysis with SGD revealed that 60% of all statistically upregulated genes (651 out of 1077) are implicated in nitrogen compound metabolism, and that genes functioning in gene expression comprise up to 49.6% (Fig. 3A). Many upregulated genes are found to be implicated in ribosome biogenesis (26.4%), RNA processing (24.8%), translation (20%) or translational initiation (2.5%). Among 285 upregulated genes in the ribosome biogenesis category (Fig. 3A), 268 genes are found to be functioning for rRNA processing. Genes involved in single organism metabolic process comprise up to 25.3% of all statistically downregulated genes (195 out of 768) (Fig. 3B).

Figure 3.

Differentially expressed genes with AgNPs. Differentially expressed genes with a q-value lesser than 0.05 were analyzed to find Gene Ontology terms (GO terms) that correspond with each individual gene's biological process. Out of 7126 total genes, 1845 were found to be statistically significant. (A) The quantification of upregulated genes associated with its specific GO term(s). Of the 1845 statistically significant genes, 1077 are found to be upregulated. (B) The quantification of downregulated genes associated with specific GO terms. The other 768 statistically significant genes were found to be downregulated.

Table 1.

Number of genes differentially expressed with 5 μg/mL AgNPs.

Total number of genes found (including noncoding genes)	Not differentially expressed	Number of differentially expressed genes (q < 0.05)	Number of differentially expressed genes (fold change ≥ 2.82)
7126	5281	1845	196
Percent of total	74.11%	25.89%	2.75%

To understand how cellular mRNA level changes upon the treatment of AgNPs, we selected 144 most upregulated and 144 most downregulated genes, a total of 288 genes. Consistent with the data shown in Fig. 3A, the vast majority of highly upregulated genes are implicated in the following, but not limited to, cellular processes: rRNA processing, ribosome biogenesis, nuclear export, rRNA transcription, response to chemicals, protein targeting, chromatin organization, DNA replication, cellular response to DNA damage stimulus, DNA repair and response to oxidative stress (Fig. 4A and Table 2). In the ‘rRNA processing’ category, DBP2 and FAF1 genes are 7.9-fold and 7.2-fold upregulated. However, the most significantly upregulated gene is CTR1, a copper transporter that mediates nearly all copper uptake under low copper conditions.

Figure 4.

GO term analysis of genes whose expression levels are significantly altered. A total of 144 most upregulated and downregulated genes with q-values less than 0.05 were analyzed and matched with corresponding GO terms to better illustrate the biological processes most affected. (A) The quantification of the 144 most upregulated genes associated to their specific GO term(s). (B) The quantification of the 144 most downregulated genes associated to their specific GO term(s).

Figure 5.

Assessment of fold change of gene expression in real-time RT-PCR experiments. Fold changes of two up and downregulated genes, which were identified by our RNAseq experiments, were compared to a housekeeping gene (ALG9) that was found to not be differentially expressed when treated with AgNPs. Fold changes were calculated with data obtained from RT-qPCR and the Pfaffl equation. The fold changes found with RT-qPCR were done to validate RNA-Seq data. (A) The calculated fold changes of the upregulated genes FAF1 and SDA1. (B) The calculated fold changes of the downregulated genes DAN1 and TIR1. Student’s t-test results are indicated either ** (P < 0.01) or *** (P < 0.001).

Table 2.

GO term analysis with 144 most upregulated genes.

Gene Ontology term	No. of genes	Corresponding genes
rRNA processing	81	BMS1, BMT5, BMT6, CGR1, DBP2, DBP3, DBP8, DBP9, DHR2, DIM1, DRS1, EBP2, ECM16, EFG1, ENP1, ENP2, ERB1, FAF1, FAL1, HAS1, HCA4, IMP4, KRI1, KRR1, MAK16, MAK5, MDN1, MRD1, MRT4, MTR4, NAN1, NIP7, NOC3, NOC4, NOG1, NOP1, NOP12, NOP14, NOP2, NOP4, NOP56, NOP58, NOP7, NOP8, NSA2, NSR1, NUG1, PRP43, PWP1, PWP2, RCL1, REX4, RIX1, RLP7, RNT1, ROK1, RPF1, RPF2, RRP12, RRP3, RRP36, RRP5, RRP8, RRS1, SAS10, SPB1, SSF1, TSR1, TSR4, URB1, UTP10, UTP11, UTP13, UTP14, UTP20, UTP21, UTP23, UTP4, UTP5, UTP6, UTP8
Ribosomal small subunit biogenesis	46	BMS1, DBP8, DHR2, DIM1, ECM16, EFG1, ENP1, ENP2, FAF1, FAL1, HAS1, IMP4, KRE33, KRI1, KRR1, LTV1, MRD1, NAN1, NOC4, NOP14, NOP58, NOP7, NSR1, PRP43, PWP2, RCL1, ROK1, RRP12, RRP3, RRP36, RRP5, RRS1, SAS10, TSR1, TSR4, UTP10, UTP11, UTP13, UTP14, UTP20, UTP21, UTP23, UTP4, UTP5, UTP6, UTP8
Ribosomal large subunit biogenesis	44	BRX1, DBP3, DBP9, DRS1, ERB1, HAS1, MAK16, MAK21, MAK5, MDN1, MRT4, NIP7, NOC2, NOG1, NOP12, NOP15, NOP2, NOP4, NOP7, NOP8, NSA1, NSA2, NUG1, PRP43, PUF6, REI1, REX4, RIX1, RIX7, RLP24, RLP7, RPF1, RPF2, RRP5, RRP8, RRS1, RSA4, SDA1, SPB1, SSF1, SYO1, URB1, YTM1, YVH1
Nuclear transport	19	ARX1, ENP1, KAP123, LTV1, MTR4, NMD3, NOG1, NOG2, NUG1, REI1, RIX1, RIX7, RPF1, RRS1, RTP1, SDA1, SRP40, SYO1, UTP8
RNA modification	16	BMT5, BMT6, DIM1, DUS3, ELP3, GAR1, GCD10, NOP1, NOP2, NOP56, PPM2, RRP8, SPB1, TRM1, TRM11, TRM2
Ribosome assembly	15	BRX1, DRS1, MAK21, MDN1, MRD1, MRT4, NSR1, REX4, RIX1, RPF1, RPF2, RRP5, RSA4, SSF1, YVH1
Organelle assembly	15	BRX1, DRS1, MAK21, MDN1, MRD1, MRT4, NSR1, REX4, RIX1, RPF1, RPF2, RRP5, RSA4, SSF1, YVH1
Transcription from RNA polymerase I promoter	11	NAN1, RPA135, RPA190, RPA34, RPA43, RPA49, RRN11, UTP10, UTP4, UTP5, UTP8
Ribosomal subunit export from nucleus	11	ARX1, LTV1, NMD3, NOG1, NOG2, NUG1, RIX1, RIX7, RPF1, RRS1, SDA1
Biological process unknown	10	CMS1, GFD2, IMD4, NOP13, NRP1, RRT14, YBL028C, YCR016W, YDL050C, YPR123C
tRNA processing	7	DUS3, ELP3, GCD10, PPM2, TRM1, TRM11, TRM2
Protein alkylation	4	EFM4, FPR4, NOP1, RMT2
Nucleobase-containing compound transport	4	ENP1, LTV1, MTR4, UTP8
Ion transport	4	AGP1, CTR1, FRE1, LTV1
RNA catabolic process	4	DBP2, MRT4, MTR4, RNT1
Peptidyl-amino acid modification	4	EFM4, FPR4, LIA1, RMT2
Cytoskeleton organization	3	LIA1, NOP15, SDA1
Response to chemical	3	ACL4, EFG1, LTV1
Transcription from RNA polymerase II promoter	3	ELP3, RNT1, TOD6
Protein targeting	3	KAP123, RTP1, SYO1
Chromatin organization	3	FPR4, NOP1, RNT1
Nucleobase-containing small molecule metabolic process	3	GUA1, RKI1, URA7
DNA replication	3	NOC3, NOP7, RIX1
snoRNA processing	3	MTR4, NOP1, RNT1
Cell wall organization or biogenesis	2	RNT1, YVH1
Regulation of organelle organization	2	FPR4, YVH1
Conjugation	2	EFG1, SSF1
Mitotic cell cycle	2	REI1, SDA1
Histone modification	2	FPR4, NOP1
Cofactor metabolic process	2	MIS1, RKI1
DNA-templated transcription, elongation	2	RPA34, RPA49
Protein dephosphorylation	2	PPT1, YVH1
Vitamin metabolic process	2	MIS1, RKI1
Organelle fission	2	EBP2, YVH1
Signaling	2	EFG1, YVH1
Cytokinesis	1	NOP15
mRNA processing	1	PRP43
Cellular response to DNA damage stimulus	1	TRM2
Vacuole organization	1	YVH1
Regulation of cell cycle	1	SDA1
RNA splicing	1	PRP43
Regulation of DNA metabolic process	1	RIX1
DNA-templated transcription, initiation	1	RRN11
Sporulation	1	YVH1
DNA-templated transcription, termination	1	RNT1
DNA repair	1	TRM2
Transmembrane transport	1	AGP1
Regulation of protein modification process	1	FPR4
Proteolysis involved in cellular protein catabolic process	1	ACL4
Pseudohyphal growth	1	KAP123
Lipid metabolic process	1	URA7
Cytoplasmic translation	1	RBG1
Response to oxidative stress	1	LTV1
Cell budding	1	REI1
Meiotic cell cycle	1	YVH1
Amino acid transport	1	AGP1
Response to osmotic stress	1	LTV1
Regulation of translation	1	PUF6
Response to starvation	1	RBG1

Our GO term analysis with 144 most downregulated genes displayed that 38 out of 144 genes are unknown in their function (Fig. 4B and Table 3). More than a dozen genes (15 out of 144) are identified to be involved in lipid metabolic processes, including genes implicated in ergosterol synthesis (ERG11, ERG25, ERG28, ERG3, ERG5, and ERG6). Among 12 downregulated transmembrane transporters, 2 genes, AAC3 and MPC3, are implicated in transporting ADP/ATP at the inner membrane of mitochondria and in transporting pyruvate to mitochondria, respectively. More genes whose functions are implicated in mitochondria-mediated cellular respiration are highly downregulated. These include AAC3, ACO1 (required for TCA cycle), CIT3 (citrate synthase) and ISF1 (affecting mitochondrial function). We identified 11 genes that play a role in responding to chemicals. In particular, 4 genes (GRX6, TSA2, VHR1 and ZTA1) out of these 11 downregulated genes are involved in an oxidative stress response, while 2 genes (CIN5 and NRG2) play a role in regulating the osmotic stress response. Among 10 genes involved in cell wall organization or biogenesis, 4 genes (GIP1, GSC2, OSW2 and SPO73) function for spore cell wall formation, whereas the rest (TIP1, TIR1, TIR2, TIR3, TIR4 and DAN1) are cell wall mannoproteins. In particular, DAN1 that codes for a cell wall mannoprotein displayed 174-fold downregulation with AgNPs.

Table 3.

Go term analysis with 144 most downregulated genes.

Gene Ontology term	No. of genes	Corresponding genes
Biological process unknown	38	CMC4, ECM13, FMP23, HBN1, ICY1, LEE1, MMO1, PBI1, RTS3, SNR190, TDA4, TOS8, TBR012C, YBR056W-A, YBR201C-A, TDR182W-A, YDR535C, YER121W, YER188W, YFL051C, YGR035C, YGR066C, YGR107W, YHL045W, YHR033W, YHR210C, YJL213W, YJL215C, YLR053C, YLR108C, YLR152C, YLR342W-A, YML083C, YML131W, YMR196W, YOR032W-A, YOR387C, YOR392W
Lipid metabolic process	15	ATF2, CIT3, CYB5, ECI1, EEB1, ERG11, ERG25, ERG28, ERG3, ERG5, ERG6, PDH1, POT1, UPC2, YEH1
Transmembrane transport	12	AAC3, ADY2, BAP2, ENA1, HXT13, JEN1, MPC3, PRM6, SCR1, STL1, SUL1, ZRT1
Response to chemical	11	ATF2, CIN5, GRX6, MF(ALPHA)2, NRG2, PRR2, TSA2, UPC2, VHR1, ZNF1, ZTA1
Monocarboxylic acid metabolic process	11	ALD4, ALD6, CIT2, CIT3, DLD3, ECI1, EEB1, FMS1, MLS1, PDH1, POT1
Ion transport	11	ADY2, ALP1, ATO2, BAP2, DIP5, ENA1, JEN1, MPC3, PRM6, SUL1, ZRT1
Cell wall organization or biogenesis	10	GIP1, GSC2, OSW2, SPO73, TIP1, TIR1, TIR2, TIR3, TIR4, DAN1
Transcription from RNA polymerase II promoter	9	CIN5, NRG2, PHD1, PRR2, ROX1, UPC2, VHR1, YAP6, ZNF1
Ergosterol biosynthesis/organization	8	HES1, ERG11, ERG25, ERG28, ERG3, ERG5, ERG6, UPC2
Carbohydrate metabolic process	8	CIT2, GSC2, GUT2, IMA1, MAL32, MLS1, PYC1, SUC2
Meiotic cell cycle	7	GIP1, GSC2, MPC54, MSH5, OSW2, SPO20, SPO73
Generation of precursor metabolites and energy	6	AAC3, ACO1, ATF2, CIT3, ISF1, RGI2
Sporulation	6	GIP1, GSC2, MPC54, OSW2, SPO20, SPO73
Transposition	6	YBL005W-B, YBL100W-B, YBR012W-B, YDR261W-B, YMR045C, YNL284C-B
Cofactor metabolic process	5	ALD4, ALD6, FMS1, GUT2, HEM13
Response to osmotic stress	5	ALD6, CIN5, ENA1, NRG2, ROX1
Cellular respiration	4	AAC3, ACO1, CIT3, ISF1
Carbohydrate transport	4	HXT13, HXT17, HXT2, STL1
Cellular amino acid metabolic process	4	BAT2, CAR2, CIT2, PUT1
rRNA processing	4	SNR10, SNR17A, SNR34, SNR37
Response to oxidative stress	4	GRX6, TSA2, VHR1, ZTA1
Protein folding	2	EUG1, HSP26
RNA modification	3	SNR10, SNR34, SNR37
Oligosaccharide metabolic process	3	IMA1, MAL32, SUC2
Membrane trafficking and protein targeting	3	COS4, SCR1, SPL2
Nucleobase-containing small molecule metabolic process	2	ALD4, ALD6, GUT2
DNA repair/recombination	2	IRC4, MSH5
Vitamin metabolic process	3	FMS1, SNZ1, THI4
Amino acid transport	3	ALP1, BAP2, DIP5
Response to starvation	3	ENA1, UPC2, VHR1
mRNA processing	2	AI1, SNR19
Mitochondrion organization	2	ACO1, THI4
Mitochondrial translation	2	15S_RRNA, 21S_RRNA
Ribosomal small subunit biogenesis	2	SNR10, SNR17A
Conjugation	2	MF(ALPHA)2, PRR2
Enzyme	2	ERR3, MAN2
Pseudohyphal growth	3	NRG2, PHD1, MIT1
Cellular response to DNA damage stimulus	1	IRC4
Translational elongation	1	ANB1
Protein dephosphorylation	1	GIP1
Protein phosphorylation	1	PRR2
Invasive growth in response to glucose limitation	1	NRG2
Signaling	1	MF(ALPHA)2

Validation of RNAseq data by RT-qPCR

To validate the gene fold-change data obtained through a RNAseq method, a real-time RT-qPCR test was used. Two upregulated genes (FAF1 and SDA1) functioning in rRNA processing/ribosome biogenesis and two downregulated genes (DAN1 and TIR1) coding for cell wall mannoproteins were chosen as well as ALG9, a housekeeping gene whose expression does not change with AgNP treatment based on our RNAseq data. The genes FAF1 and SDA1 exhibited 2.88 ± 0.44-fold and 3.12 ± 0.53-fold upregulation in gene expression with treatment of 5 μg/mL AgNPs, whereas DAN1 and TIR1 expression levels decrease more than 270-fold and 12-fold, respectively (Fig, 5A&B). Together, the results of gene fold change measurement with RT-qPCR are consistent with our RNAseq assay.

Determination of ROS in cells incubated with AgNPs

Several genes implicated in mitochondrial functions are downregulated (Fig. 4B and Table 3), and therefore, we wanted to test levels of ROS originated from the mitochondria. To measure the total level of ROS from the mitochondria we used DHR123, which enters the cell and is oxidized by ROS to form R123 and to emit green fluorescence (Kiani-Esfahani et al.2012). It was found that total levels of mitochondria-driven ROS in response to up to 10 μg/mL of AgNPs was not changed when compared to the ROS level of non-treated groups or 5 μg/mL AgNP-treated groups (Fig. 6A–D). Given that the concentration of superoxide in cells with stress rises, we then determined levels of superoxide by using DHE, a superoxide indicator emitting red fluorescence (Nazarewicz, Bikineyeva and Dikalov 2013). Interestingly, our data showed that levels of superoxide in the presence of both 5 and 10 μg/mL of AgNP were decreased (Fig. 6E–H). It is not clear why superoxide levels in cells with AgNPs decreases, but Jones et al. (2011) recently proposed that cellular superoxide can be quenched in the presence of Ag ions derived from AgNPs. Therefore, we concluded that the decrease of superoxide in AgNP-treated cells may be due to ionized Ag reacting with superoxide in the cell.

Figure 6.

Assessment of ROS and superoxide levels in flow cytometry experiments. Two mitochondrial ROS indicators, DHR123 and DHE, were utilized in the quantification of ROS and superoxide in yeast cells treated with varying concentrations of 20 nm AgNPs for 8 h. DHR123 and DHE were added at concentrations of 5 μg/mL for the last 2 h of incubation. Each treatment concentration (0, 5 and 10 μg/mL of AgNPs) was tested in triplicate. A two-tailed equal variance Student's t-test was performed, and no statistical difference was observed between the three groups. (A–C) Representative DHR123 fluorescent intensity charts of cells treated with DHR123 and grown in the presence of no AgNPs (A), 5 μg/mL AgNPs (B) or 10 μg/mL AgNPs. (D) Quantification of DHR123 levels. The total % fluorescence means from each replicate in the ROS detection assay. (E-G) Representative DHE fluorescent intensity charts of cells treated with DHE and grown in the presence of no AgNPs (E), 5 μg/mL AgNPs (F) or 10 μg/mL AgNPs (G). (H) The total % fluorescence means from each replicate in the superoxide detection assay.

Figure 7.

Assessment of cell wall stability with Zymolase 100T enzyme. The rate of yeast cell wall degradation was observed in samples treated with 0, 5 and 10 μg/mL AgNPs, and each sample was tested in triplicate. When treated with Zymolase 100T, the OD’s of each sample were recorded every 10 min for 6 h. The rate of cell wall degradation mediated by Zymolase 100T indicates the integrity of cell walls when incubated with varying concentrations of AgNPs. The empty circles represent the mean of the samples treated with no Zymolase 100T. The solid black squares represent the mean of the samples treated with 0 μg/mL AgNPs and 10 μg/mL Zymolase 100T. The empty diamonds represent the mean of the samples treated with 5 μg/mL AgNPs and 10 μg/mL Zymolase 100T. The solid black triangles represent the mean of the samples treated with 10 μg/mL AgNPs and 10 μg/mL Zymolase 100T.

Figure 8.

Schematic model of changes of cellular processes with spherical 20 nm AgNPs in yeast cells. (A) AgNPs appear to affect the integrity of ribosome, which might end up elevating expression levels of genes implicated in rRNA processing and the biogenesis of small large subunit ribosomes as well as nuclear export of ribosomes. (B) Several classes of cellular activities appear to be downregulated by the presence of AgNPs, including cell wall/membrane integrity, sugar import, metabolism in the cytosol, cellular respiration in mitochondria and protein folding.

Investigation of cell wall integrity in cells incubated with AgNPs

We have found several genes involved in cell wall organization and genes that code for cell wall structural proteins, such as mannoproteins, to be downregulated (Fig. 4). We wanted to test the integrity of the cell wall in yeast treated with varying concentrations of AgNPs. To investigate the stability of the cell wall, we used an enzyme, Zymolase 100T, that creates spheroplasts and degrades the cell wall. It was found that the rates of cell wall degradation in samples treated with Zymolase 100T were much higher in samples treated with 10 μg/mL AgNPs than 5 μg/mL AgNPs and the same relationship was observed between 5 μg/mL AgNPs and samples not treated with AgNPs (Figure 7). Therefore, we concluded that the rates of cell wall degradation of AgNP-treated cells are higher than non-treated cells and the cell walls of AgNP-treated cells are less stable than their untreated counterparts.

DISCUSSION

There are increasing concerns for AgNPs’ potential environmental risks due to the fact that AgNPs are widely used in many commercial products. However, the temporal resolution of their effects on cellular and molecular dynamics is poorly understood. We elucidated the molecular mechanisms of cytotoxicity caused by AgNPs in the budding yeast, S. cerevisiae, by comprehensively investigating global mRNA expression patterns, and to our knowledge the present study is the first RNAseq report that indicates that a sublethal amount of AgNPs negatively affects many cellular processes occurring in the budding yeast. It is interesting to note that AgNO₃ also inhibited yeast cell proliferation in a dose-dependent manner (Supplementary Figure 1). Though the current study does not illustrate differentially expressed genes, a recent study revealed a significant overlap (13–21%) of differentially expressed genes among AgNO₃- and AgNP-treated Arabidopsis thaliana (Kaveh et al.2013), indicating that this gene alteration in AgNP-treated groups was, at least in part, originated from Ag ions released by AgNPs. In the future, it is of great interest to identify differentially expressed genes shared (or distinctive) upon exposure to AgNPs and Ag ions.

Upregulated mRNAs and their potential impacts on the cell integrity

We found that more than 80 genes out of 144 most upregulated genes upon 5 μg/mL treatment of AgNPs are identified to function for rRNA processing/ribosome biogenesis. Many translated products of these 80 genes locate at the nucleolus, associated with rRNA processing/ribosome biogenesis (Fig. 6A). For instance, genes coding for Enp2, Faf1 and its binding partner Krr1 (Zheng et al.2014) are highly elevated in their expression with AgNP treatment. All of these proteins are synthesized in the cytoplasm and then travel to the nucleolus to help with 18S rRNA processing, a component of the small ribosomal subunit (Zheng et al.2014). According to a previously published paper from Dr. Baserga lab, Krr1 is a component of the small subunit (SSU) processome, a 2.2-MDa ribonucleoprotein complex involved in the processing, assembly and maturation of the SSU of the ribosome in Eukaryotes (Phipps, Charette and Baserga 2011). Upon looking closely at the ‘rRNA processing’ and ‘ribosome small subunit biogenesis’ rows in Table 2, we were able to find 22 more upregulated genes that code for the known and putative protein components of the yeast SSU processome: Nop1, Nop56, Nop58, Imp4, Utp10, Utp13, Utp21, Rrp36, Utp11, Utp14, Noc4, Utp20, Utp23, Utp24, Bms1, Dbp8, Dhr2, Rc1, Rok1, Rrp3, Rrp5, Enp1. Our upregulated gene list (Table 2) also contains 39 genes that code for assembly factors that function in maturation of 60S ribosomal subunit (or large subunit [LSU] of ribosome) in S. cerevisiae (Woolford and Baserga 2013), including Rrp5, Nop4, Urb1, Nop8, Mak21, Noc2, Dbp3, Mak5, Ssf1, Mak16, Brx1, Rpf1, Ytm1, Erb1, Nop7, Drs1, Nop15, Nsa1, Rlp7, Has1, Nop2, Rpf2, Rrs1, Rlp24, Nog1, Spb1, Nsa2, Nug1, Rix7, Rix1, Mdn1, Rsa1, Sda1, Rei1, Yvh1, Mrt4, Puf6, Arx1 and Nmd3. Consistent with these upregulations of factors for SSU and LSU of ribosome, a number of genes that code for RNA polymerase 1 holoenzyme were upregulated upon AgNP treatment (See ‘transcription from RNA polymerase 1 promoter’ category of Table 2). Tying together, our observation is that sublethal amounts of AgNPs in the budding yeast stimulate ribosome biogenesis at the nucleolus (Figure 8A). A fundamental question in relation to this observation is the potential cause of these upregulations. Previous results from Escherichia coli–AgNP interaction studies can provide hints for interpreting our RNAseq data. It has been shown that AgNPs release Ag⁺ ions, which bind to thiol groups (SH) of the protein (Klueh et al.2000; Rai et al.2012). Furthermore, AgNPs were found to interact with ribosomes in a manner similar to the binding mode suggested above, leading to denaturation or inactivation of ribosome proteins and thereby resulting in inhibition of translation and protein synthesis (Morones et al.2005; Jung et al.2008; Rai et al.2012). By interpolating the negative consequence of Ag⁺ binding to ribosomes into our RNAseq results, one can put forth the idea that Ag⁺ affects yeast ribosome functions negatively via temporal or stable interactions with ribosomal components. In response to the signal of the presence of compromised or inefficient ribosomes, the cell might stimulate de novo formation of healthy ribosomes, for which the corresponding activities of SSU and LSU processosomes, rRNA synthesis, tRNA synthesis (see Table 3 for upregulated genes implicated in tRNA processing) and ribosome export to the cytoplasm (Table 3) should be upregulated (see our model in Figure 8A).

Downregulated mRNAs and their potential impacts on the cell integrity

It has been well known that AgNPs display their antimicrobial potential through impairing biological membranes. Ag⁺ ions released from the surface of AgNPs can bind to proteins carrying SO₄⁻, causing irreversible structural alteration, which in turn disrupts cell membrane integrity (Ghosh et al.2012; Abbaszadegan et al.2015). In agreement to this concept, a recent SEM (scanning electron microscopy) study with Candida albicans revealed that yeasts with AgNPs 0.0089 ppm treatment for 1 day displayed a rough cell surface, indicating outer cell wall damage (Lara et al.2015). Their TEM (transmission electron microscopy) data further displayed that the yeast cell wall treated with AgNPs was swollen thicker (nearly 2-fold increase in thickness) and was partially disruptive, consistent with the presence of holes and pits on the cell wall of yeast treated with AgNPs (Kim et al.2009). Importantly, our RNAseq results in conjunction with GO term analysis provide a more comprehensive view of defects in cell wall organization caused by AgNPs. First, it appears that AgNPs affect proper turnover rate of cell wall components including mannoprotein and glycans. We postulate this idea because the thickening of yeast cell wall upon treatment of AgNPs (Lara et al.2015) might be due to either a unsteady organization of the cell wall or an accumulation of cell wall proteins. The idea of destabilization of the cell wall has been proven true (Kim et al.2009), while the possibility of the accumulation of cell wall mannoproteins and glucans in the presence of AgNPs has yet to be tested. Based on our RNAseq data, the following seven genes implicated in cell wall organization are highly downregulated: TIR1–4, DAN1, GSC2 and RNT1. These four TIR and DAN1 genes code for cell wall mannoproteins, while the gene product of GSC2 is a catalytic subunit of β-1,3-glucan synthase required for cell wall glucan synthesis and remodeling in S. cerevisiae (Lesage and Bussey 2006). RNT1 gene is involved in cell wall stress response and in regulating degradation of cell wall integrity (Catala, Aksouh and Abou Elela 2012). The significant reduction of these mRNA levels indicates the possibility that the cell senses an abnormally thick cell wall caused by AgNPs. Even though the thicker wall does not necessarily mean an increase in the number of mannoproteins and glucans, we cannot exclude the possibility that the abnormality is, in part, due to an accumulation of these cell wall components. In this scenario, genes for these cell wall components can be downregulated since the regulatory factors for cell wall integrity would act against the expression of cell wall components (Fig. 8B, model B).

In light of the finding of the downregulation of six ergosterol synthesis genes (Table 3), including ERG11, 25, 28, 3, 5 and 6, one can conjecture that the integrity of the plasma membrane is compromised in the presence of AgNPs. Consistent with this idea, genes for membrane sugar transporters such as HXT13, HXT17 and HXT2 are differentially expressed (Table 3 and Fig. 8B, model B). Furthermore, multiple genes implicated in carbohydrate metabolism in the cytoplasm (Fig. 6, model B) were downregulated. The possibility is that both destabilization of the plasma membrane and decreased levels of sugar transporters with AgNPs limit the amount of sugar to be metabolized in the cytosol (Fig. 8B, model B). In addition to the metabolic defects, genes associated with pyruvate transport (MPC3), TCA cycle (ACO1, CIT2, CIT3) and NADH regeneration (ALD4 and GUT2) were downregulated with AgNPs (Table 3, and Fig. 8B). Therefore, our observation is not only consistent with previous findings that demonstrated that nanosilver particles cause direct mitochondrial damage, disturb the function of respiratory chain, increase ROS production and induce apoptosis (Hsin et al.2008; AshaRani et al.2009; Bressan et al.2013), but also provide new mechanistic insights into mitochondrial dysfunction mediated by AgNPs. In addition, a recent study with S. cerevisiae where 9 nm AgNPs were used showed that 5 or 10 μg/mL of AgNPs caused a drastic inhibition of cellular respiration taking place in mitochondria (Galvan Marquez et al.2018), augmenting the notion that mitochondrial functions are downregulated. Consistently, our present study shows downregulation of AAC3, a ADP/ATP translocator gene functioning for exchanging ADP generated by the F₁F₀-ATPase for ATP (Table 3, not shown in Fig. 8B) (Kolarov, Kolarova and Nelson 1990; Giraud and Velours 1997). It is highly likely that low levels ADP in the mitochondrial matrix due to downregulation of AAC3 by AGNPs limit the substrate concentration for the ATP synthase, which may limit the amount of ATP production. Yet to maintain cell viability, the limited amount of ATP must be exported to the cytoplasm by abnormally low levels of Aac3. A decrease of intracellular ATP levels would negatively affect uncountable biological processes including EUG1 and HSP26 gene product-mediated protein folding, regardless of their action locations (Fig. 8B).

CONCLUSION

In the present study, we assessed potential toxicity of AgNPs and provided evidence that yeast cells exposed to these NPs displayed minor defect in growth rates. Accordingly, the presence of 5 μg/mL AgNPs in the culture media led to significant transcriptome changes in yeast, manifested by the differential expression of several hundred genes implicated in diverse cellular processes. Given many genes that play roles in ribosome biogenesis, cell wall/membrane integrity and mitochondrial functions are significantly altered with the treatment of AgNPs, our conclusion is that even sublethal amount of AgNPs could serve as a potential environmental stress factor to living cells.

Acknowledgements

We are indebted to Dr. Laszlo Kovacs for generously providing TruSeq Stranded mRNA LT Sample Preparation Kit (Illumina). We thank Rishi Patel for organizing regular face-to-face meetings and providing relevant input and feedback on this project.

FUNDING

This work was funded by the U.S. Army Engineer Research and Development Center—Environmental Laboratory through the Environmental Quality and Technology Program, contract no. W912HZ-15-2-0032 P00002.

Conflict of interest. None declared.