A combined proteomics and metabolomics approach ...

Nanotoxicology

ISSN: 1743-5390 (Print) 1743-5404 (Online) Journal homepage: http://www.tandfonline.com/loi/inan20

A combined proteomics and metabolomics approach to assess the effects of gold nanoparticles in vitro. Sabrina Gioria, Joana Lobo Vicente, Paola Barboro, Rita La Spina, Giorgio Tomasi, Patricia Urbán, Agnieszka Kinsner-Ovaskainen, François Rossi & Hubert Chassaigne To cite this article: Sabrina Gioria, Joana Lobo Vicente, Paola Barboro, Rita La Spina, Giorgio Tomasi, Patricia Urbán, Agnieszka Kinsner-Ovaskainen, François Rossi & Hubert Chassaigne (2015): A combined proteomics and metabolomics approach to assess the effects of gold nanoparticles in vitro., Nanotoxicology, DOI: 10.3109/17435390.2015.1121412 To link to this article: http://dx.doi.org/10.3109/17435390.2015.1121412

View supplementary material

Accepted author version posted online: 09 Dec 2015.

Submit your article to this journal

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=inan20 Download by: [European Commission]

Date: 10 December 2015, At: 01:31

Just Accepted by Nanotoxicology A combined proteomics and metabolomics approach to assess the effects of gold nanoparticles in vitro. Sabrina Gioria, Joana Lobo Vicente, Paola Barboro, Rita La Spina, Giorgio Tomasi, Patricia Urbán, Agnieszka Kinsner-Ovaskainen, François Rossi and Hubert Chassaigne doi: 10.3109/17435390.2015.1121412

Downloaded by [European Commission] at 01:31 10 December 2015

Abstract Omics technologies, such as proteomics or metabolomics, have to date been applied in the field of nanomaterial safety assessment to a limited extent. To address this dearth, we developed an integrated approach combining the two techniques to study the effects of two sizes, 5 and 30 nm, of gold nanoparticles (AuNPs) in Caco-2 cells. We observed differences in cells exposed for 72 h to each size of AuNPs: 61 responsive (up/down-regulated) proteins were identified and 35 metabolites in the cell extract tentatively annotated. Several altered biological pathways were highlighted by integrating the obtained multi-omics data with bioinformatic tools. This provided a unique set of molecular information on the effects of nanomaterials at cellular level. This information was supported by complementary data obtained by immunochemistry, microscopic analysis and multiplexed assays. A part from increasing our knowledge on how the cellular processes and pathways are affected by nanomaterials (NMs), these findings could be used to identify specific biomarkers of toxicity or to support the safe-by-design concept in the development of new nanomedicines.

© 2015 Taylor & Francis. This provisional PDF corresponds to the article as it appeared upon acceptance. Fully formatted PDF and full text (HTML) versions will be made available soon. DISCLAIMER: The ideas and opinions expressed in the journal’s Just Accepted articles do not necessarily reflect those of Taylor & Francis (the Publisher), the Editors or the journal. The Publisher does not assume any responsibility for any injury and/or damage to persons or property arising from or related to any use of the material contained in these articles. The reader is advised to check the appropriate medical literature and the product information currently provided by the manufacturer of each drug to be administered to verify the dosages, the method and duration of administration, and contraindications. It is the responsibility of the treating physician or other health care professional, relying on his or her independent experience and knowledge of the patient, to determine drug dosages and the best treatment for the patient. Just Accepted articles have undergone full scientific review but none of the additional editorial preparation, such as copyediting, typesetting, and proofreading, as have articles published in the traditional manner. There may, therefore, be errors in Just Accepted articles that will be corrected in the final print and final online version of the article. Any use of the Just Accepted articles is subject to the express understanding that the papers have not yet gone through the full quality control process prior to publication.

A combined proteomics and metabolomics approach to assess the effects of gold nanoparticles in vitro.

Sabrina Gioria1*, Joana Lobo Vicente1, Paola Barboro2, Rita La Spina 1, Giorgio Tomasi1, 1, Patricia Urbán1, Agnieszka Kinsner-Ovaskainen1, François Rossi1 and Hubert Chassaigne1


1

European Commission, Joint Research Centre, Institute for Health and Consumer Protection, Via Enrico Fermi

2749, I-21027 Ispra, Italy 2

IRCCS Azienda Ospedaliera Universitaria San Martino - IST Istituto Nazionale per la Ricerca sul Cancro,

Largo Rosanna Benzi 10, I-16132 Genova, Italy

*Corresponding author at: Joint Research Centre, IHCP, NBS Unit, TP125, via E. Fermi 2749, 21027 Ispra, VA, Italy. Tel.: +39 0332783584; Fax: +39 0332 785787. E-mail address: [email protected] (S. Gioria)

1

Currently working for the University of Copenhagen, Nørregade 10, DK-1017 Copenhagen, Denmark

Abstract

Omics technologies, such as proteomics or metabolomics, have to date been applied in the field of nanomaterial safety assessment to a limited extent. To address this dearth, we developed an integrated approach combining the two techniques to study the effects of two sizes, 5 and 30 nm, of gold nanoparticles (AuNPs) in Caco-2 cells. We observed differences in cells exposed for 72 h to each size of AuNPs: 61 responsive (up/down-regulated) proteins


were identified and 35 metabolites in the cell extract tentatively annotated. Several altered biological pathways were highlighted by integrating the obtained multi-omics data with bioinformatic tools. This provided a unique set of molecular information on the effects of nanomaterials at cellular level. This information was supported by complementary data obtained by immunochemistry, microscopic analysis and multiplexed assays. A part from increasing our knowledge on how the cellular processes and pathways are affected by nanomaterials (NMs), these findings could be used to identify specific biomarkers of toxicity or to support the safe-by-design concept in the development of new nanomedicines.

Keywords: Two-dimensional gel electrophoresis (2DE), liquid chromatography highresolution tandem mass spectrometry (LC-HRMS/MS), omics data treatment, systems biology analysis.

Introduction

Currently, omics techniques are routinely applied as high-throughput methods in medical diagnostics for the analysis of disease states as well as in basic research to study biological processes or in predictive toxicology. Among the different omics techniques, genomics, which gives an overview of the complete set of genetic information provided by the DNA, and transcriptomics, which looks into gene expression patterns, convey only limited


information about phenotyping, therefore are of restricted value and are considered as an entry point with respect to the other more recent omics sciences. By comparison, the information acquired by proteomic studies, based on the identification of de-regulated dynamic protein products and their interactions, have a higher scientific value. Metabolomics, the newest omics science, which refers to the analysis of the complete set of low molecular weight compounds in a sample, has the great advantage to provide a closer link to the phenotype of a cell/tissue at a specified time under particular environmental conditions.

To date, post genomics techniques have been applied in the field of nanomaterial risk assessment to a limited extent (Jia et al., 2013, Kobeissy et al., 2014). Furthermore, the integration of multi-omics data has only recently been performed (Bartel et al., 2015, Eichner et al., 2014, Decourcelle et al., 2015, Meierhofer et al., 2014, Kutmon et al., 2014, Fagerberg et al., 2014, Diez et al., 2015, Cooney et al., 2015). The proteome and metabolome are directly interconnected as protein levels influence the metabolic profile of a cell system and metabolites' concentration may affect protein expression. Therefore, an integrated approach, that combines proteomics and metabolomics data, is a very powerful tool to provide a more comprehensive understanding of biological effects of potential toxicants, including NMs.

Although there is an emerging interest in applying metabolomics to nanotoxicology (Huang et al., 2012, Lu et al., 2011) and in its integration with molecular profiling at protein level

(Eichner et al., 2014), further research in these fields is needed. The proteomic approach which is quite standardised, is not yet routinely implemented in the field of nanotoxicology whereas the metabolomic approach is complex and still at its early stage of development. This is not only due to the chemical diversity of cellular metabolites but also to the redundancy of cellular metabolic pathways that make data elaboration and interpretation challenging.


In this contribution, we present a multi-omics approach to better understand the cellular effects after AuNPs exposure in vitro. As progress in metabolomics data processing and analysis lags behind that of proteomics, with this work we attempt to provide useful tools to improve metabolomics data processing and to integrate multi-omics data at systems biology level. In more detail, we investigated the modification that occurs in the proteome and the metabolome profile of human colon adenocarcinoma cells (Caco-2) (Sambuy et al., 2005) when exposed to AuNPs. Qualitative and quantitative data of de-regulated proteins and metabolites obtained using two-dimensional gel electrophoresis (2DE) and liquid chromatography high-resolution tandem mass spectrometry (LC-HRMS/MS) were combined and interpreted using systems biology analysis. The aim of this study was, apart from demonstrating the great potential of this approach, to contribute to the understanding of the biological processes that are affected when AuNPs interact with living systems. We focused on AuNPs, as they have attracted in the last decade enormous scientific and technological interest due to their unique chemical properties, which are tuneable, by changing the size, shape or surface chemistry. The initial claim of absence of cytotoxicity (Connor et al., 2005) has raised enthusiasm, and led to an increased use of AuNPs in consumer products such as cosmetics, food packaging, toothpastes, food supplements and

lubricants (Sung et al., 2011), as well as in medicine for cell imaging (Chen et al., 2005), targeted drug delivery (Ghosh et al., 2008), cancer diagnostics and therapy (Paciotti et al., 2004, Huang et al., 2007). However, recent studies demonstrated size dependent cytotoxicity induced by AuNPs in vitro (Coradeghini et al., 2013, Vetten et al., 2013, Alkilany and Murphy, 2010). Therefore, this work aims at contributing to a better understanding of the cellular and


molecular mechanisms triggered by exposure to AuNPs in Caco-2 cells by applying a multiomics approach cross-linked with a broad set of techniques to support the omics findings. Integration of proteomics and metabolomics data is crucial to obtain biological insight for a correct hazard assessment. Furthermore, information achieved based on the knowledge of networks, processes and pathways modified, can be used to improve drug design or to identify specific biomarkers of toxicity.

Methods A schematic of the experimental design and the analytical and bioinformatics tools employed is provided in Figure 1. Due to the complexity of the comprehensive combined approach used in this study, the detailed methodology on AuNPs synthesis and characterization, the 2D gelbased proteomics analysis, the MS-based metabolomics and the bioinformatics description


for the systems biology analysis can be found in Supplementary Material.

In vitro experiments Cell culture conditions Human colon adenocarcinoma Caco-2 cells (Sigma-Aldrich®, Milano, Italy) were maintained in complete culture medium composed of Dulbecco's Modified Eagle Medium (DMEM) high glucose (4500 mg/L) (Lifetechnologies, Italy) supplemented with 10% (v/v) Fetal Bovine Serum

(FBS,

North

America

Origin,

Lifetechnologies,

Italy),

0.5%

(v/v)

penicilin/streptomycine, 4 mM L-glutamine and 1% (v/v) not essential amino acids (Lifetechnologies, Italy). For routine culture cells were maintained in a sub-confluent state under standard cell culture conditions in a humidified incubator (37 °C, 5% CO2, 95% humidity) (Heraeus Thermo Fisher®, Belgium).

AuNPs treatment and sample preparation For proteomic experiments, 1x106 Caco-2 cells were seeded in 5 mL complete culture medium in 100 x 20 mm Petri dish (Corning, Italy). After 24 h, the medium was replaced and 5 or 30 nm AuNPs was added to obtain the final concentrations of 300 µM (59 µg/mL). In each experiment, untreated cells were used as control. Six biological replicates were

performed for each experimental condition. Proteins extraction from the cytoplasmatic compartment was performed after 72 h of exposure time as described in (Gioria et al., 2014). For metabolomics experiments, cells were prepared as described above. At the end of the 72 h exposure time, the cell culture medium was removed. Cells were washed with 5 mL of cold phosphate-buffered saline solution (PBS) (Lifetechnologies, Italy) and the wash solution discarded. Cell lysates were obtained by adding 500 µL of ice-cold methanol to each well and mechanically harvested with a sterile plastic disposable cell scraper. The lysate was


transferred in a 1.5 mL Eppendorf® tube. Each dish was then washed with an additional 250 µL ice-cold methanol that was collected into the respective Eppendorf® tube. Recovered cell lysate was sonicated at 50 W for 5 min and further centrifuged at 15,000 g for 15 min at 4 °C. The supernatant was collected and stored in a new 1.5 mL Eppendorf® tube at -80 °C. The methanol solution was evaporated to dryness using the centrifugal vacuum evaporator (Univapo 150 ECH, Uniequip Germany) for 30 min with a cooling system at 10 °C. The samples were re-suspend in 100 µL of the LC-MS mobile phase (0.1% formic acid (FA) in a solution of milli-Q water: Methanol, 95:5) and centrifuged at 15,000 x g for 10 min at 4 °C. Samples were transferred into 96-well plates and then covered with a suitable cover mat for LC-HRMS analyses.

2D gel-based proteomic experiments The 2D gel-based proteomics analyses together with the proteomic data processing have been performed according to our previous work (Gioria et al., 2014). Minor changes have been done and can be found in Supplementary Methods.

MS-based metabolomic experiments In this work, we refer to the reporting standards in metabolomics (http://msiworkgroups.sourceforge.net) of the Metabolomics Society (Sumner et al., 2007). These standards recommend that authors should report the level of identification for all metabolites base on a four-level system ranging from level 1 (identified compounds), via levels 2 and 3 (putatively-annotated compounds) and 3 (putatively-compound classes) to level 4 (unidentified or unclassified metabolites which nevertheless can be differentiated based upon


spectral data). Going through the identification of all metabolites of interest was limited to a small number of cases as the identification challenge is immense and confident unambiguous assignments of observed metabolic features to a single compound are not always possible (Warwick B. Dunn et al., 2013). Definitive (level 1) identification would require the comparison of two or more orthogonal properties (eg. rt, accurately measured m/z, fragmentation mass spectrum) of a chemical standard to the same properties observed for the metabolite of interest analysed under identical analytical conditions. In this work, a level of confidence of 2 for metabolite identification (putatively annotated compounds) was reached. The application of accurate measurement of m/z was able to provide putative HMDB annotations (top 5 possibilities were considered). Details concerning metabolomics analysis and data processing are available in Supplementary Methods.

Systems biology analysis Relation between the identified proteins and annotated metabolites was evaluated using the software Ingenuity Pathways Analysis (IPA) (Ingenuity Systems ®, Redwood City, CA, USA). For more details refer to Supplementary Methods.

Other techniques used For Trypan Blue assay, immunocytochemistry analysis and apoptosis array membrane refer to Supplementary Methods.

Results Identification of the differentially expressed proteins


We assessed the differences in the cytoplasmic proteome of Caco-2 cells by 2DE, exposed to 5 or 30 nm AuNPs at a concentration of 300 μM with respect to the control (untreated) cells. The analysis of imaged gels identified a total of 836 (range 722 – 907), 819 (range 777 – 867) and 760 (range 574 – 944) protein spots for the control, 5 or 30 nm AuNPs treated samples, respectively (Figure 2), with an average percentage of matched spots across gels of 94%. The Mann-Whitney test was used to compare the overall protein expression profile. Among the protein spots detected, 18 were found to be down-regulated by at least 2-fold, and 18 were up-regulated by at least 2-fold in cells treated with 5 nm AuNPs. In the case of 30 nm AuNPs treatment, 28 spots were down-regulated and 5 spots were up-regulated with at least a twofold change. Only 33% of the de-regulated proteins were found to be common to both treatments. Differences were also found between AuNPs treated cells, as 23 spots were differentially expressed in cells treated with 5 nm AuNPs with respect to 30 nm AuNPs (17 up-regulated and 6 down-regulated) A total number of 66 proteins were differentially expressed and significant (p < 0.05) between the two groups (5 and 30 nm AuNPs) among which 6 were common to the two comparisons (Figure 2). The differences in protein deregulation observed between the two sizes can be explained considering the uptake. Indeed, as shown by our group in an earlier work, the number of internalized NPs by the cell is much

higher in the case of the smaller 5 nm size compared to the bigger size 30 nm AuNPs (Bajak et al., 2014). Proteins of interest (up- or down-regulated) were defined from the images of the control and treated samples, and their corresponding spots in the image of the preparative gel were matched. The 66 spots were selected from the preparative gel for spot picking. The aforementioned de-regulated protein targets were identified using LC-MS/MS. Proteome


Discoverer (Thermo Scientific®) with the Sequest workflow and UniProtKB/Swiss-Prot database were used for protein identification of the selected de-regulated protein spots. Results were returned for 61 out of the 66 selected spots (Supplementary Material – Figure S1). Table 1 reports the spot and code number, the UniProt Accession Number, the protein code, the p value and fold change as compared to the control, the protein coverage, the theoretical MW and pI and the protein score. In the table, proteins that have been differentially regulated by AuNPs are classified according to their main function based on UniProtKB/Swiss-Prot and Gene Ontology (GO). In figure 3A proteins deregulated by 5 or 30 nm AuNPs are visualised grouped for main function.

Investigation of de-regulated metabolites Using LC-MS and metabolomic data analysis, we assessed the differences in the metabolic profile within the cells when exposed to AuNPs of 5 or 30 nm sizes in respect to the untreated control. The m/z features obtained after statistical analysis were submitted to public database search for metabolite annotations. For metabolite identification, we used the standard reporting system, described previously (Sumner et al., 2007). Results show that both sizes of AuNPs had a significant impact on the metabolic profiling of the cell extract as 35 metabolites were found significantly differentially expressed (p 90%) in all conditions. We investigated the effects of AuNPs exposure on the cytoskeleton


organization and nuclei by immunocytochemistry, staining the F-actin filaments of the cells with phalloidin and nuclei with DAPI. In the absence of NPs, the actin filaments were well organized, whereas cells exposed to AuNPs presented an alteration in F-actin distribution, with thinner and poorly oriented stress fibers and a more punctuate fluorescent pattern (Figure 3S-B). This modification of the cytoskeleton might alter the shape, growth rate, and survival mechanisms of the cells. Fragmentation of the cell nuclei was also observed after exposure to AuNPs (Figure 3S-A). In order to quantify these microscopic observations, INCell Analyzer was used to study the nuclei and cytoskeleton morphology of Caco-2 cells. Results obtained (Figure 4S) confirmed that the cytoskeleton of the cells incubated with AuNPs was significantly disrupted (p < 0.05), although no differences were observed in the F-actin analysis between 5 nm and 30 nm treatments (p > 0.05). The nuclei area of the cells was significantly reduced after exposure to AuNPs (p < 0.05), and several fragmented nuclei were detected which are linked to apoptotic cell death. Furthermore, the nuclear measurements applied as an apoptosis indicator showed significant increased values in both NPs treatments with respect to the control (p < 0.05). Both parameters related to the nuclei, area and apoptosis indicator, reveal apoptotic events in NP-treated cells with a more pronounced effect in 5 nm AuNPs (5 nm vs 30 nm, p < 0.05).

To support these findings, cells were stained with Annexin V antibody (Figure 3S-C) and a simultaneous screening of 43 human markers was performed using an apoptosis-specific antibody array. The results shown in Figure 5S highlight significant changes in the expression of several proteins involved in the apoptosis process. CAS-3 and CAS-8 were found strongly upregulated by exposure to 5 nm AuNPs. In addition, BAD, BAX, BCL-2, BCL-W, CIAP-2, BID, BIM, CD40, CYTOC, DR6, FAS, several proteins of the IGFBP family, sTNF-R2 and


TNF-ɑ were found down-regulated by 30 nm AuNPs.

Discussion Current challenges in omics approach Proteomics and metabolomics, combined with systems biology, are excellent tools to screen for effects and toxicological mechanisms induced by NPs. However, the metabolomics approach remains very challenging and still relatively novel in the field of nanotoxicology. In the present work, results from metabolomics were integrated to data obtained from the study of the de-regulation of the proteome to closely investigate the bio-responses of Caco-2 cells when exposed to AuNPs. To partially overcome the metabolomics data analysis challenge, IPA was used to integrate both proteomics and metabolomics data and to gain insights into molecular pathways from identified proteins and metabolite annotations. This combined approach, allowed the detection of significant changes in the proteome and metabolome profiles of Caco-2 cells treated with 5 or 30 nm AuNPs. Due to the complexity of the experimental set-up, the number of conditions tested was limited to two sizes of AuNPs and one dose (300 µM). The dose was selected based on previous cytotoxicity tests (Bajak et al. 2014) and corresponds to the lowest one where significant cytotoxicity started to be observed in at least one of the two NPs sizes tested (5 nm). Unfortunately, at this molar concentration of gold, it was not possible to conduct exposure experiments using 30 nm

particles while maintaining the same total surface area or number of particles as the experiments with 5 nm particles. To achieve these conditions, based on the same starting mass concentration of gold in the stock solution, it would have been necessary to preconcentrate the 30 nm nanoparticles solution to unreachable levels (36 times more for equivalent surface area and 216 for equivalent particle number). Such concentration would cause the agglomeration of the NPs. For this reason, the exposure has been normalized to gold molarity.


The limitations of 2D gel-based proteomics platform are the association of isoelectricfocusing (IEF) and two-dimensional electrophoresis as quantitative preparative techniques for mass spectrometry, hence only a reduced number of samples can be processed. However, the attractiveness of this platform is that it allows the identification of post-translational modifications that occurred to proteins, otherwise not detected with a direct mass spectrometry based-approach. By 2DE, we were able to observe isoforms (spot G5 and F5, A1 and B3, E9-E10-E11) as well as proteolytic fragments (D1 and D2). In addition, we combined the advanced omics tools with a broad set of complementary techniques, including immunochemistry, microscopic analysis and multiplexed assays for the detection of apoptosis proteins. Cross-linking different techniques can improve the knowledge and understanding in nanotoxicology as shown in this work.

Mechanisms of AuNPS toxicity A number of intrinsic cellular pathways are triggered in response to AuNPs exposure. Using IPA, the differentially expressed proteins and metabolites were used to show a range of molecular

networks

modulated

by

AuNPs.

These

include

cellular

compromise

(degeneration), small molecule biochemistry, cell morphology, cellular assembly and organization, cellular growth and proliferation (Table 3 and Figure 4).

Changes in the Caco-2 cells' proteome and metabolome due to AuNPs exposure are sizespecific, with 5 nm AuNPs inducing a more severe protein de-regulation compared to the 30 nm particles. That can be explained considering the much higher internalization of 5 nm AuNPs when compared to the 30 nm AuNPs one (Bajak et al., 2014). For both AuNPs sizes, down-regulated proteins were mainly associated with cellular growth and proliferation, whereas the proteins found mostly up-regulated were involved in antioxidant activity and apoptosis. This was particularly evident in the case of exposure to 5 nm AuNPs.


Interestingly, the proteins which were most strongly de-regulated by 5 and 30 nm AuNPs were ACTBL2 that is involved in cytoskeleton organization, HOOK2 responsible for vesicle trafficking from early endosome to late endosome, PSAT1 and ACAT1 involved in amino acid biosynthesis and ketone metabolism respectively. G3BP1 involved in cellular response to stress (Tourriere et al., 2003) was under-expressed in both AuNPs treatments. On the contrary, AHCY was over-expressed by both AuNPs treatment. AHCY belongs to the adenosylhomocysteinase

family

and

catalyses

the

reversible

hydrolysis

of

S-

adenosylhomocysteine (AdoHcy) to adenosine (Ado) and L-homocysteine (Hcy), thus regulating the intracellular S-adenosylhomocysteine (SAH) concentration. In this regard, it has been shown that SAH can promote apoptosis, inhibit migration and adhesion (Li et al., 2014).

Cell viability, oxidative stress and survival mechanisms Various, well-established in vitro assays to study cellular DNA damage, inflammation, oxidative stress and mitochondrial injury have been largely used as measurement end points for assessing NP-induced toxicity. However, the information obtained from these assays is not sufficient to provide a detailed description of the overall bio-responses nor to highlight the critical biochemical pathways which are affected by NPs exposure. On the other hand

omics technologies, such as proteomics and metabolomics, provide more mechanistic information. Among the differentially expressed proteins found to be de-regulated in our study, several are associated with cell viability and survival mechanisms, including oxidative stress and apoptosis. We have reported overexpression of VIL-1 in 5 nm AuNPs exposed cells, a protein that plays a key role in actin regulation, in the organization, integration and regulation of multiple


epithelial cell functions such as cell morphology, cell motility and appears to regulate cell death by maintaining mitochondrial integrity. PCK2, ACAA2 and ARHGEF7 were found to be up-regulated by the smaller AuNPs, whereas P4HB and HSPD1 resulted down-regulated by 5 nm AuNPs. At this regards, it has been shown that PCK2 is over-expressed under stress condition, ARHGEF7 function as a positive regulator of apoptosis, whereas ACAA2 abolishes BNIP3-mediated apoptosis and mitochondrial damage (Mendez-Lucas et al., 2014). P4HB has been reported to be involved in the regulation of intrinsic apoptotic signalling pathway in response to oxidative stress. Recent studies (Lee et al., 2014) have shown that P4HB is among the down-regulated proteins involved in the endoplasmic reticulum stress response and Nrf2-ARE signalling. HSPD1 is implicated in mitochondrial protein import and macromolecular assembly. It also facilitates the correct folding of imported proteins, preventing misfolding and promoting the refolding and proper assembly of unfolded polypeptides generated under stress conditions in the mitochondrial matrix. UBA1, that has a critical importance in regulation of diverse cellular processes such as cell cycle and cell death, was found down-regulated in cells exposed to 30 nm AuNPs. In addition, expression of several proteins with oxido-reductase function was modified in cells exposed to 5 nm AuNPs (IDH1, MDH2, SDH4, ETFA) and 30 nm AuNPs (LDHA).

Among the differentially expressed annotated metabolites, propionylcarnitine (C-3 carnitine) and glycine levels were found increased after exposure to both AuNP sizes. These metabolites have already been reported as associated to the apoptotic processes (Halama et al., 2011, Ferrara et al., 2005, Ankarcrona et al., 1995). The integration of proteomics and metabolomics data clearly confirmed that proteins and related metabolites involved in carbohydrate metabolism and in stress response were affected by AuNPs treatments. The decrease of glycolytic rate can be correlated to the growth rate reduction. Moreover,


metabolomics analysis showed accumulation of glutathione (GSH) in both 5 and 30 nm treated cells, which is considered a natural self-defence mechanism of cells to cope with oxidative stress since GSH plays an essential role in maintaining the intracellular redox environment. In addition, for 5 nm AuNPs exposure we noticed a decrease in the stress response proteins HSPD1 and HSPA9, which function as chaperons. It is known that in mammalian cells, chaperones present anti-apoptotic activity by preventing caspase activation (Longo et al., 2015). Of interest, is also the down-regulation observed of YWHAG known for its capacity to protect cells against stress-induced apoptosis. Thus, the lower abundance of these proteins reflects the apoptotic state of cells that underwent AuNPs treatment, more enhanced in the 5 nm AuNPs treated cells. To further confirm that apoptosis occurred in Caco-2 cells exposed to AuNPs, the omics data were supported by florescence microscopy analysis where over-expression of Annexin V and nuclear fragmentation induced by AuNPs were evident and more pronounced in the case of 5 nm AuNPs. The findings were supported by data obtained using an apoptosis-specific antibody array. Results show that exposure to 30 nm AuNPs reduces the expression of antiapoptotic BCL-2 family members, of CIAP-2 that is required to protect cells against TNF-ɑ induced apoptosis, of several proteins of the IGFBP family and of sTNF-R2. Several antiand pro-apoptotic proteins are triggered by 30 nm AuNPs, although we suggest that there is a

balance among these elements which prevents a massive activation of Caspase-3 (cas-3) in Caco-2 cells exposed to AuNPs. Conversely, the effects of 5 nm AuNPs are more severe, thus cas-3 and cas-8 expression was found to be significantly up-regulated. The established link between metabolism and apoptosis that we report here represents a novelty in the area of NPinduced toxicity. We suggest that the metabolomics signatures may be used as early biomarkers of apoptosis in in vitro systems exposed to NPs. Cell cycle, cellular growth and proliferation


Our results reveal that the differences in cell growth already evident by morphological analysis are reflected at protein and metabolite level. YWHAG involved in G2/M transition of mitotic cell cycle, MCM7 and NUDC were all down-regulated in cells exposed to 5 nm AuNPs. In addition, we report changes in cellular amino acid levels. Exposure to both AuNPs sizes result in a decrease in amino acids such as L-leucine and L-isoleucine; whereas glycine level is increased. As glycine is biosynthesized starting from the amino acid serine, its change in the metabolite profiles in treated cells is perfectly in accordance with the increased expression of PSAT1, a phosphoserine aminotransferase (Vié et al., 2008) involved in serine biosynthesis.

Cytoskeleton organisation and cell adhesion We show here that several proteins involved in cytoskeleton organization and cell adhesion are de-regulated by AuNPs exposure. We reported that both sizes of AuNPs trigger the upregulation of ACTBL2, a key protein in orchestrating cytoskeletal reorganization and cell migration. In addition, CTTN, which contributes to the organization of the actin cytoskeleton and cell structure, together with CCT2, known to play a role in vitro in the folding of actin and tubulin, were de-regulated after 5 nm AuNPs exposure. Furthermore, of relevance is the de-

regulation of VIL-1caused by the smaller AuNPs size. This protein which functions in the capping, severing, and bundling of actin filaments, is a dominant component of the brush border cytoskeleton. The exposure to the larger AuNPs decreases the expression of TMOD3 which blocks the elongation and depolymerization of the actin filaments at the pointed end. We also report a reduction in the expression of several proteins involved in cell adhesion. CTNND1 that was


found strongly down-regulated after 5 nm NPs exposure, is associated with, and regulates the cell adhesion properties of C-, E- and N-cadherins being therefore critical for the surface stability. In this regard, we also noted a decrease of the level of DSG1, a member of the cadherin cell adhesion molecule superfamily, and of JUP, a member of the catenin protein family involved in regulating the coordination of cell-cell adhesion, following exposure to 5 and 30 nm AuNPs, respectively. Our work points out that exposure to AuNPs induces a severe alteration at cytoskeleton and cell adhesion level. In addition, the major proteins involved in these modifications were highlighted.

DNA synthesis and repair, protein synthesis, and amino acid transport Moreover, the proteomic data show that several proteins involved in DNA synthesis and repair, protein synthesis, and amino acid transport were also found de-regulated. At protein level, MCM7 and GSPT1 are down-regulated following exposure to 5 nm and 30 nm AuNPs, respectively. On the other hand, EEF1D was up-regulated in cells exposed to 30 nm AuNPs. Among the de-regulated metabolites, glutathione, a major endogenous antioxidant produced by the cells, participating directly in the neutralization of free radicals and reactive oxygen compounds and having also a role in metabolic and biochemical reactions such as DNA

synthesis and repair, protein synthesis, and amino acid transport, decreased in Caco-2 cells exposed to both AuNPs treatment. In this respect, changes in glutathione metabolism level were already observed at gene expression level (Bajak et al., 2014). The expression of two proteins, ATIC and PFAS, involved in purine biosynthesis, are found strongly decreased in cells exposed to 30 nm AuNPs.

Cell interaction and intracellular distribution of AuNPs


The modifications observed at protein and metabolite level are clearly due to intracellular accumulation of AuNPs (Bajak et al., 2014, Gioria et al., 2014). In relation to this, it has been shown that the uptake of AuNPs into mammalian cells occurs more likely via endocytosis (Huefner et al., 2014). We report that SH3GL1 and EAA1 proteins, both known to be involved in endocytic transport, were found down-regulated in 5 nm exposed cells. In this respect, (Renard et al., 2014) have shown the important role of SH3GL1 in clathrinindependent endocytic process. In addition, ANX2 was up-regulated in 30 nm AuNPs exposed cells and it was shown that ANX2 binds to endosomes and functions in endosomal transport (Emans et al., 1993), a process that is regulated by tyrosine 23 phosphorylation (Morel and Gruenberg, 2009). Taken together, the proteomic data not only suggests that the intracellular uptake of AuNPs in Caco-2 cells is mediated by endocytosis, but allow the identification of key–player proteins involved in this mechanism.

Conclusions Recent advances in proteomics and untargeted metabolomics technologies enable qualitative and quantitative analysis of a wide range of biomolecules and thus allow the determination of their relative abundance between different biological conditions. Here, we provide a welldesigned, robust and reliable methodology to evaluate changes in the protein and metabolite profiles in vitro, thus contributing to the progress in assessing NP-induced toxicity. In addition, our work also aims at bringing metabolomics to a similar level as that of the


proteomics approach by working on the workflow for data processing as well as on the harmonisation of the process. We have shown that the integration of omics technologies together with complementary methods offers not only a promising tool to understand the toxicological behaviour of nanomaterials, but could also enhance nano-drug development and allow the identification of biomarkers for NPs-induced toxicity. Future work aims at highlighting similarity among the de-regulated biomolecules in different cell models exposed to diverse NPs.

Funding The research described in this work was supported by the European Commission Joint Research Centre (JRC) within the Nanobiosciences (NBS) and Chemical Assessment and Testing (CAT) Units of the Institute for Health and Consumer Protection (IHCP) through the JRC Multiannual Work Programme.

Acknowledgements We are very thankful to Dr. Martin Hajduch, Dr. Claude Guillou, Dr. Douglas Gilliland and Dr. Jessica Ponti for helpful discussions. We would like to acknowledge Dr. Claudia Placidi and Dr. Giovanna Finzi, of the University of Insubria, Varese, Italy for their assistance. We thank Dr. Val Millar (GE Healthcare Life Sciences) for technical and scientific support in the


development of analysis protocols for INCell Analyzer.

Conflict of interest The authors declare that there are no conflicts of interest.


References ALKILANY, A. M. & MURPHY, C. J. 2010. Toxicity and cellular uptake of gold nanoparticles: what we have learned so far? J Nanopart Res, 12, 2313-2333. ANKARCRONA, M., DYPBUKT, J. M., BONFOCO, E., ZHIVOTOVSKY, B., ORRENIUS, S., LIPTON, S. A. & NICOTERA, P. 1995. Glutamate-induced neuronal death: a succession of necrosis or apoptosis depending on mitochondrial function. Neuron, 15, 961-73. BAJAK, E., FABBRI, M., PONTI, J., GIORIA, S., OJEA-JIMENEZ, I., COLLOTTA, A., MARIANI, V., GILLILAND, D., ROSSI, F. & GRIBALDO, L. 2014. Changes in Caco-2 cells transcriptome profiles upon exposure to gold nanoparticles. Toxicol Lett, 233, 187-199. BARTEL, J., KRUMSIEK, J., SCHRAMM, K., ADAMSKI, J., GIEGER, C., HERDER, C., CARSTENSEN, M., PETERS, A., RATHMANN, W., RODEN, M., STRAUCH, K., SUHRE, K., KASTENMULLER, G., PROKISCH, H. & THEIS, F. J. 2015. The Human Blood MetabolomeTranscriptome Interface. PLoS Genet, 11, e1005274. CHEN, J., SAEKI, F., WILEY, B. J., CANG, H., COBB, M. J., LI, Z. Y., AU, L., ZHANG, H., KIMMEY, M. B., LI, X. & XIA, Y. 2005. Gold nanocages: bioconjugation and their potential use as optical imaging contrast agents. Nano Lett, 5, 473-7. CONNOR, E. E., MWAMUKA, J., GOLE, A., MURPHY, C. J. & WYATT, M. D. 2005. Gold nanoparticles are taken up by human cells but do not cause acute cytotoxicity. Small, 1, 325-7. COONEY, J. M., BARNETT, M. P., DOMMELS, Y. E., BREWSTER, D., BUTTS, C. A., MCNABB, W. C., LAING, W. A. & ROY, N. C. 2015. A combined omics approach to evaluate the effects of dietary curcumin on colon inflammation in the Mdr1a mouse model of inflammatory bowel disease. J Nutr Biochem. CORADEGHINI, R., GIORIA, S., GARCIA, C. P., NATIVO, P., FRANCHINI, F., GILLILAND, D., PONTI, J. & ROSSI, F. 2013. Size-dependent toxicity and cell interaction mechanisms of gold nanoparticles on mouse fibroblasts. Toxicol Lett, 217, 205-16. DECOURCELLE, M., PEREZ-FONS, L., BAULANDE, S., STEIGER, S., COUVELARD, L., HEM, S., ZHU, C., CAPELL, T., CHRISTOU, P., FRASER, P. & SANDMANN, G. 2015. Combined transcript, proteome, and metabolite analysis of transgenic maize seeds engineered for enhanced carotenoid synthesis reveals pleotropic effects in core metabolism. J Exp Bot, 66, 3141-50. DIEZ, P., DROSTE, C., DEGANO, R. M., GONZALEZ-MUNOZ, M., IBARROLA, N., PEREZANDRES, M., GARIN-MUGA, A., SEGURA, V., MARKO-VARGA, G., LABAER, J., ORFAO, A., CORRALES, F. J., DE LAS RIVAS, J. & FUENTES, M. 2015. Integration of Proteomics and Transcriptomics Data Sets for the Analysis of a Lymphoma B-Cell Line in the Context of the Chromosome-Centric Human Proteome Project. J Proteome Res, 14, 3530-40. EICHNER, J., ROSENBAUM, L., WRZODEK, C., HARING, H. U., ZELL, A. & LEHMANN, R. 2014. Integrated enrichment analysis and pathway-centered visualization of metabolomics, proteomics, transcriptomics, and genomics data by using the InCroMAP software. J Chromatogr B Analyt Technol Biomed Life Sci, 966, 77-82. EMANS, N., GORVEL, J. P., WALTER, C., GERKE, V., KELLNER, R., GRIFFITHS, G. & GRUENBERG, J. 1993. Annexin II is a major component of fusogenic endosomal vesicles. J Cell Biol, 120, 1357-69. FAGERBERG, L., HALLSTROM, B. M., OKSVOLD, P., KAMPF, C., DJUREINOVIC, D., ODEBERG, J., HABUKA, M., TAHMASEBPOOR, S., DANIELSSON, A., EDLUND, K., ASPLUND, A., SJOSTEDT, E., LUNDBERG, E., SZIGYARTO, C. A., SKOGS, M., TAKANEN, J. O., BERLING, H., TEGEL, H., MULDER, J., NILSSON, P., SCHWENK, J. M., LINDSKOG, C., DANIELSSON, F., MARDINOGLU, A., SIVERTSSON, A., VON FEILITZEN, K., FORSBERG, M., ZWAHLEN, M., OLSSON, I., NAVANI, S., HUSS, M., NIELSEN, J., PONTEN, F. & UHLEN, M. 2014. Analysis of the human tissue-specific expression by genome-wide integration of transcriptomics and antibody-based proteomics. Mol Cell Proteomics, 13, 397-406. FERRARA, F., BERTELLI, A. & FALCHI, M. 2005. Evaluation of carnitine, acetylcarnitine and isovalerylcarnitine on immune function and apoptosis. Drugs Exp Clin Res, 31, 109-14. GHOSH, P., HAN, G., DE, M., KIM, C. K. & ROTELLO, V. M. 2008. Gold nanoparticles in delivery applications. Adv Drug Deliv Rev, 60, 1307-15.


GIORIA, S., CHASSAIGNE, H., CARPI, D., PARRACINO, A., MESCHINI, S., BARBORO, P. & ROSSI, F. 2014. A proteomic approach to investigate AuNPs effects in Balb/3T3 cells. Toxicol Lett, 228, 111-26. HALAMA, A., MOLLER, G. & ADAMSKI, J. 2011. Metabolic signatures in apoptotic human cancer cell lines. OMICS, 15, 325-35. HUANG, S. M., ZUO, X., LI, J. J., LI, S. F., BAY, B. H. & ONG, C. N. 2012. Metabolomics studies show dose-dependent toxicity induced by SiO(2) nanoparticles in MRC-5 human fetal lung fibroblasts. Adv Healthc Mater, 1, 779-84. HUANG, X., JAIN, P. K., EL-SAYED, I. H. & EL-SAYED, M. A. 2007. Gold nanoparticles: interesting optical properties and recent applications in cancer diagnostics and therapy. Nanomedicine (Lond), 2, 681-93. HUEFNER, A., SEPTIADI, D., WILTS, B. D., PATEL, II, KUAN, W. L., FRAGNIERE, A., BARKER, R. A. & MAHAJAN, S. 2014. Gold nanoparticles explore cells: cellular uptake and their use as intracellular probes. Methods, 68, 354-63. JIA, L., LU, Y., SHAO, J., LIANG, X. J. & XU, Y. 2013. Nanoproteomics: a new sprout from emerging links between nanotechnology and proteomics. Trends Biotechnol, 31, 99-107. JOHN TURKEVICH, P. C. S., JAMES HILLIER 1951. A study of the nucleation and growth processes in the synthesis of colloidal gold. Discussions of the Faraday Society, 11, 20. KOBEISSY, F. H., GULBAKAN, B., ALAWIEH, A., KARAM, P., ZHANG, Z., GUINGABCAGMAT, J. D., MONDELLO, S., TAN, W., ANAGLI, J. & WANG, K. 2014. Post-Genomics Nanotechnology Is Gaining Momentum: Nanoproteomics and Applications in Life Sciences. OMICS, 18, 111-131. KUHL, C., TAUTENHAHN, R., BOTTCHER, C., LARSON, T. R. & NEUMANN, S. 2012. CAMERA: an integrated strategy for compound spectra extraction and annotation of liquid chromatography/mass spectrometry data sets. Anal Chem, 84, 283-9. KUTMON, M., EVELO, C. T. & COORT, S. L. 2014. A network biology workflow to study transcriptomics data of the diabetic liver. BMC Genomics, 15, 971. LEE, L. C., WENG, Y. T., WU, Y. R., SOONG, B. W., TSENG, Y. C., CHEN, C. M. & LEE-CHEN, G. J. 2014. Downregulation of proteins involved in the endoplasmic reticulum stress response and Nrf2-ARE signaling in lymphoblastoid cells of spinocerebellar ataxia type 17. J Neural Transm, 121, 601-10. LI, Q., MAO, L., WANG, R., ZHU, L. & XUE, L. 2014. Overexpression of S-adenosylhomocysteine hydrolase (SAHH) in esophageal squamous cell carcinoma (ESCC) cell lines: effects on apoptosis, migration and adhesion of cells. Mol Biol Rep, 41, 2409-17. LIU, X., ATWATER, M., WANG, J. & HUO, Q. 2007. Extinction coefficient of gold nanoparticles with different sizes and different capping ligands. Colloids Surf B Biointerfaces, 58, 3-7. LONGO, V., ZDRALEVIC, M., GUARAGNELLA, N., GIANNATTASIO, S., ZOLLA, L. & TIMPERIO, A. M. 2015. Proteome and metabolome profiling of wild-type and YCA1-knock-out yeast cells during acetic acid-induced programmed cell death. J Proteomics, 128, 173-188. LU, X., TIAN, Y., ZHAO, Q., JIN, T., XIAO, S. & FAN, X. 2011. Integrated metabonomics analysis of the size-response relationship of silica nanoparticles-induced toxicity in mice. Nanotechnology, 22, 055101. MEIERHOFER, D., WEIDNER, C. & SAUER, S. 2014. Integrative analysis of transcriptomics, proteomics, and metabolomics data of white adipose and liver tissue of high-fat diet and rosiglitazonetreated insulin-resistant mice identified pathway alterations and molecular hubs. J Proteome Res, 13, 5592-602. MENDEZ-LUCAS, A., HYROSSOVA, P., NOVELLASDEMUNT, L., VINALS, F. & PERALES, J. C. 2014. Mitochondrial phosphoenolpyruvate carboxykinase (PEPCK-M) is a pro-survival, endoplasmic reticulum (ER) stress response gene involved in tumor cell adaptation to nutrient availability. J Biol Chem, 289, 22090-102. MOREL, E. & GRUENBERG, J. 2009. Annexin A2 binding to endosomes and functions in endosomal transport are regulated by tyrosine 23 phosphorylation. J Biol Chem, 284, 1604-11. PACIOTTI, G. F., MYER, L., WEINREICH, D., GOIA, D., PAVEL, N., MCLAUGHLIN, R. E. & TAMARKIN, L. 2004. Colloidal gold: a novel nanoparticle vector for tumor directed drug delivery. Drug Deliv, 11, 169-83.


PRINCE, J. T. & MARCOTTE, E. M. 2006. Chromatographic alignment of ESI-LC-MS proteomics data sets by ordered bijective interpolated warping. Anal Chem, 78, 6140-52. RENARD, H. F., SIMUNOVIC, M., LEMIERE, J., BOUCROT, E., GARCIA-CASTILLO, M. D., ARUMUGAM, S., CHAMBON, V., LAMAZE, C., WUNDER, C., KENWORTHY, A. K., SCHMIDT, A. A., MCMAHON, H. T., SYKES, C., BASSEREAU, P. & JOHANNES, L. 2014. Endophilin-A2 functions in membrane scission in clathrin-independent endocytosis. Nature. SAMBUY, Y., DE ANGELIS, I., RANALDI, G., SCARINO, M. L., STAMMATI, A. & ZUCCO, F. 2005. The Caco-2 cell line as a model of the intestinal barrier: influence of cell and culture-related factors on Caco-2 cell functional characteristics. Cell Biol Toxicol, 21, 1-26. SMITH, C. A., WANT, E. J., O'MAILLE, G., ABAGYAN, R. & SIUZDAK, G. 2006. XCMS: processing mass spectrometry data for metabolite profiling using nonlinear peak alignment, matching, and identification. Anal Chem, 78, 779-87. SUMNER, L. W., AMBERG, A., BARRETT, D., BEALE, M. H., BEGER, R., DAYKIN, C. A., FAN, T. W., FIEHN, O., GOODACRE, R., GRIFFIN, J. L., HANKEMEIER, T., HARDY, N., HARNLY, J., HIGASHI, R., KOPKA, J., LANE, A. N., LINDON, J. C., MARRIOTT, P., NICHOLLS, A. W., REILY, M. D., THADEN, J. J. & VIANT, M. R. 2007. Proposed minimum reporting standards for chemical analysis Chemical Analysis Working Group (CAWG) Metabolomics Standards Initiative (MSI). Metabolomics, 3, 211-221. SUNG, J. H., JI, J. H., PARK, J. D., SONG, M. Y., SONG, K. S., RYU, H. R., YOON, J. U., JEON, K. S., JEONG, J., HAN, B. S., CHUNG, Y. H., CHANG, H. K., LEE, J. H., KIM, D. W., KELMAN, B. J. & YU, I. J. 2011. Subchronic inhalation toxicity of gold nanoparticles. Part Fibre Toxicol, 8, 16. TOURRIERE, H., CHEBLI, K., ZEKRI, L., COURSELAUD, B., BLANCHARD, J. M., BERTRAND, E. & TAZI, J. 2003. The RasGAP-associated endoribonuclease G3BP assembles stress granules. J Cell Biol, 160, 823-31. VETTEN, M. A., TLOTLENG, N., TANNER RASCHER, D., SKEPU, A., KETER, F. K., BOODHIA, K., KOEKEMOER, L. A., ANDRAOS, C., TSHIKHUDO, R. & GULUMIAN, M. 2013. Label-free in vitro toxicity and uptake assessment of citrate stabilised gold nanoparticles in three cell lines. Part Fibre Toxicol, 10, 50. VIÉ, N., COPOIS, V., BASCOUL-MOLLEVI, C., DENIS, V., BEC, N., ROBERT, B., FRASLON, C., CONSEILLER, E., MOLINA, F., LARROQUE, C., MARTINEAU, P., DEL RIO, M. & GONGORA, C. 2008. Overexpression of phosphoserine aminotransferase PSAT1 stimulates cell growth and increases chemoresistance of colon cancer cells. Mol Cancer, 7, 14. WARWICK B. DUNN, ERBAN, A., WEBBER, R. J. M., CREEK, D. J., BROWN, M., BREITLING, R., HANKEMEIER, T., GOODACRE, T., NEUMANN, S., KOPKA, J. & VIANT, M. R. 2013. Mass appeal: metabolite identification in mass spectrometry-focused untargeted metabolomics. Metabolomics, 9, 22. WISHART, D. S., KNOX, C., GUO, A. C., EISNER, R., YOUNG, N., GAUTAM, B., HAU, D. D., PSYCHOGIOS, N., DONG, E., BOUATRA, S., MANDAL, R., SINELNIKOV, I., XIA, J., JIA, L., CRUZ, J. A., LIM, E., SOBSEY, C. A., SHRIVASTAVA, S., HUANG, P., LIU, P., FANG, L., PENG, J., FRADETTE, R., CHENG, D., TZUR, D., CLEMENTS, M., LEWIS, A., DE SOUZA, A., ZUNIGA, A., DAWE, M., XIONG, Y., CLIVE, D., GREINER, R., NAZYROVA, A., SHAYKHUTDINOV, R., LI, L., VOGEL, H. J. & FORSYTHE, I. 2009. HMDB: a knowledgebase for the human metabolome. Nucleic Acids Res, 37, D603-10.

Tables with captions Table 1. List of de-regulated proteins identified in individual 2D gel spots of Caco-2 cytoplasmic extracts. Each protein spot has been assigned a UniProt Accession Number, the protein symbol, the protein coverage, the number of identified peptides and amino acids, the molecular mass, the calculated pI and a probability score. Proteins have been classified according to their main function based on UniProtKB/Swiss-Prot and Gene Ontology (GO). Quantitative changes after AuNPs treatment (5 nm AuNPs vs Ctrl and 30 nm AuNPs vs Ctrl)


are reported (p value and fold change) for individual proteins. Log 2 = 0.58, thus the range was set from -0.58 to 0.58 and colour-coded. Green for < -0.58, 0 to black and >0.58 to red. The values in between are shown as colour gradients.


Table 1

Metabolism Amino Acids Adenosylhomocysteinase OS=Homo sapiens GN=AHCY PE=1 SV=4 - [SAHH_HUMAN]

Phosphoserine aminotransferase OS=Homo sapiens GN=PSAT1 PE=1 SV=2 - [SERC_HUMAN]

Lipid and sterol 3-ketoacyl-CoA thiolase, mitochondrial OS=Homo sapiens GN=ACAA2 PE=1 SV=2 [THIM_HUMAN]

Acetyl-CoA acetyltransferase, mitochondrial OS=Homo sapiens GN=ACAT1 PE=1 SV=1 [THIL_HUMAN]

Purine biosynthesis Phosphoribosylformylglycinamidi ne synthase OS=Homo sapiens GN=PFAS PE=1 SV=4 [PUR4_HUMAN]

Bifunctional purine biosynthesis protein PURH OS=Homo sapiens GN=ATIC PE=1 SV=3 -

S p o t

A c c C e o ss d io e n

N u m b e r

N N u u m m b b e e r r

4 3 0 7

2 2 2 2

1 3 1 2

1 2 4 1

6 9 2 1 3 5 3

P r ot ei n c o d e

P 2 3 5 2 6 Q 9 Y 6 1 7

A H C Y

A C A A 2

C 1 1

P 4 2 7 6 5 P 2 4 7 5 2

B 1 1 A 0 8

O 1 5 0 6 7 P 3 1

P F A S

F 0 3

E 0 8

F 0 8

P S A T 1

A C A T 1

A T I

p v a l u e

0 . 0 0 3 1 0 . 0 0 1 5 0 . 0 0 9 8 0 . 0 0 5 2

5nm AuNPs F ol log d 2.F ch old a ch n an ge ge

3. 01 7

58 .5 77

6. 56 7

17 .5 21

1.5 93

5.8 72

p v a l u e

0 . 0 2 7 7 0 . 0 1 7 9

30nm AuNPs F ol log d 2.F ch old a ch n an ge ge

2. 58 9

18 .6 97

1.3 73

4.2 25

19. 9/9

17. 3/6

21. 7/1 1

2.7 15

4.1 31

Co ve ra ge / un iq ue pe pti de s

0 . 0 4 3 2 0 . 0 3 4 5 0 . 0

76 .0 74

4. 34 5 35 .8

6.2 49

2.1 19 5.1 63

30. 7/9

7.5 5/9 15. 5/8

T h eo re ti ca l

E xp er im en tal

P r o t e i n

M W /p I

M W /p I

4 7. 7/ 6. 3

41 .6/ 6. 1

4 7 . 9 9

4 0. 4/ 7. 7

36 .3/ 7. 4

1 5 . 7 4

4 1. 9/ 8. 1

41 .2/ 7. 8

5 1 . 9 8

4 5. 2/ 8. 8

37 .9/ 7. 7

6 0 . 6 0

1 4 4. 6/ 5. 8 6 4. 6/

10 5. 3/ 5. 5

3 4 . 7 7

59 .1/ 7.

1 9 .

S c o r e

[PUR9_HUMAN]

Carboxilic acid Pyridoxal-dependent decarboxylase domain-containing protein 1 OS=Homo sapiens GN=PDXDC1 PE=1 SV=2 [PDXD1_HUMAN]

6

9 3 9

C

2 8 5

Q 6 P 9 9 6

P D X D C 1

0 . 0 0 3 5

L D H A

0 . 0 3 7 2 0 . 0 0 4 6

G 0 5

P 0 0 3 3 8 P 0 6 7 3 3 P 0 0 5 5 8 P 1 0 5 1 5 P 1 0 5 1 5

2 6 0 2

E 0 4

Q 1 6 8 2 2

P C K 2

3 5 2

F 0 1

P 2 9

T K T

7 7 1 9

B 0 1


Energy Glycolysis L-lactate dehydrogenase A chain OS=Homo sapiens GN=LDHA PE=1 SV=2 - [LDHA_HUMAN] 1 1 8 Alpha-enolase OS=Homo sapiens GN=ENO1 PE=1 SV=2 [ENOA_HUMAN]

Phosphoglycerate kinase 1 OS=Homo sapiens GN=PGK1 PE=1 SV=3 - [PGK1_HUMAN]

Dihydrolipoyllysine-residue acetyltransferase component of pyruvate dehydrogenase complex, mitochondrial OS=Homo sapiens GN=DLAT PE=1 SV=3 [ODP2_HUMAN] Dihydrolipoyllysine-residue acetyltransferase component of pyruvate dehydrogenase complex, mitochondrial OS=Homo sapiens GN=DLAT PE=1 SV=3 [ODP2_HUMAN] Gluconeogenesis Phosphoenolpyruvate carboxykinase [GTP], mitochondrial OS=Homo sapiens GN=PCK2 PE=1 SV=3 [PCKGM_HUMAN] Penthose phosphate pathway Transketolase OS=Homo sapiens GN=TKT PE=1 SV=3 [TKT_HUMAN]

1 3 2 0

1 2 0 9

6 6 0 1

5 6 1 5

B 0 6

C 0 2

F 0 6

F 0 5

E N O 1

P G K 1

D L A T

D L A T

0 . 0 2 9 2 0 . 0 1 5 2 0 . 0 1 5 4

8. 94 1

2. 36 2

2. 27 2

3.1 60

1.2 40

1.1 84

16

7. 58 9

5. 21 4

5. 07 6

2.9 24

2.3 82

2.3 44

18. 0/1 1

19. 0/6

37. 6/1 5

46. 3/1 8

23. 5/1 4

7.7 /4

0 . 0 1 1 4

3. 11 5

1.6 39

12. 8/7

0 . 0

3. 20 4

1.6 80

15. 9/9

6. 7

2

3 8

8 6. 7/ 5. 4

80 .7/ 5. 1

3 7 . 0 5

3 6. 7/ 8. 3

32 .0/ 8. 7

1 5 . 1 2

4 7. 1/ 7. 4

46 .2/ 7. 8

4 4. 6/ 8. 1

40 .2/ 7. 7

6 9. 0/ 7. 8

64 .6/ 5. 7

6 9. 0/ 7. 8

64 .7/ 5. 8

1 0 5 . 1 9 1 1 1 . 5 3 1 0 6 . 6 2 8 . 3 8

7 0. 7/ 7. 6

63 .7/ 7. 5

2 8 . 6 0

6 7. 8/

64 .8/ 6.

2 4 .

4

TCA pathway Isocitrate dehydrogenase [NADP] cytoplasmic OS=Homo sapiens GN=IDH1 PE=1 SV=2 [IDHC_HUMAN]


Malate dehydrogenase, mitochondrial OS=Homo sapiens GN=MDH2 PE=1 SV=3 [MDHM_HUMAN]

Aconitate hydratase, mitochondrial OS=Homo sapiens GN=ACO2 PE=1 SV=2 [ACON_HUMAN]

Succinate dehydrogenase [ubiquinone] flavoprotein subunit, mitochondrial OS=Homo sapiens GN=SDHA PE=1 SV=2 [DHSA_HUMAN] Respiration ATP synthase subunit d, mitochondrial OS=Homo sapiens GN=ATP5H PE=1 SV=3 [ATP5H_HUMAN]

E-transport Electron transfer flavoprotein subunit alpha, mitochondrial OS=Homo sapiens GN=ETFA PE=1 SV=1 - [ETFA_HUMAN]

Transcription RNA transport Leucine-rich PPR motifcontaining protein, mitochondrial OS=Homo sapiens GN=LRPPRC PE=1 SV=3 - [LPPRC_HUMAN]

Transcription regulation Far upstream element-binding protein 2 OS=Homo sapiens

4 0 1

7. 7

5

5 7

0 . 0 0 9 4 0 . 0 1 1 8 0 . 0 0 6 1 0 . 0 3 6 1

4 6. 6/ 7. 0

47 .5/ 5. 2

9 . 4 6

3 5. 5/ 8. 7

34 .1/ 7. 9

9 8 . 3 8

8 5. 4/ 7. 6

76 .7/ 7. 5

9 4 . 8 8

7 2. 6/ 7. 4

83 .4/ 6. 4

5 8 . 8 4

1 8. 5/ 5. 3

21 .5/ 5. 2

2 3 . 8 0

3 5. 1/ 8. 4

28 .8/ 7. 5

7 2 . 3 3

10 4. 4/ 5. 5

1 8 3 . 7 8

72 .5/

2 1

E 1 2

O 7 5 8 7 4 P 4 0 9 2 6 Q 9 9 7 9 8 P 3 1 0 4 0

C 1 0

O 7 5 9 4 7

A T P 5 H

G 0 4

P 1 3 8 0 4

E T F A

6 9 0 7

F 1 0

P 4 2 7 0 4

L R P P R C

0 . 0 3 9 6

17 .6 13

4.1 39

30. 3/3 5

1 5 7. 8/ 6. 1

3 6

E 0

Q 9

K H

0 .

2.

1.5

10. 0/5

7 3.

7 3 1 4

1 1 0 9

2 7 2 3

4 8 2 3

7 0 1 2

2 1 0 9

A 0 5

D 0 3

F 0 4

I D H 1

2 0 2

M D H 2

A C O 2

S D H A

13 .8 33

76 .8 02

2. 90 6

18 .4 70

3.7 90

10. 9/4

6.2 63

44. 4/1 2

20. 8/1 4

1.5 39

4.2 07

18. 8/9 0 . 0 1 5 8

0 . 0 0 1 0

3. 35 5

3. 67 2

1.8 77

43. 5/6

39. 6/1 1

1.7 46

GN=KHSRP PE=1 SV=4 [FUBP2_HUMAN]

Mediator of RNA polymerase II transcription subunit 4 OS=Homo sapiens GN=MED4 PE=1 SV=1 [MED4_HUMAN]


Ribosomal proteins 60S acidic ribosomal protein P0 OS=Homo sapiens GN=RPLP0 PE=1 SV=1 - [RLA0_HUMAN]

Protein Folding and stability 60 kDa heat shock protein, mitochondrial OS=Homo sapiens GN=HSPD1 PE=1 SV=2 [CH60_HUMAN]

60 kDa heat shock protein, mitochondrial OS=Homo sapiens GN=HSPD1 PE=1 SV=2 [CH60_HUMAN]

0 4

8 1 1 5

4 4 0 2

9 2 0 1

7 5 1 0

6

B 1 0

2 9 4 5 Q 9 N P J 6

G 0 7

P 0 5 3 8 8

R P L P 0

0 . 0 3 0 7

P 1 0 8 0 9 P 1 0 8 0 9 P 1 4 6 2 5

H S P D 1

0 . 0 3 1 3 0 . 0 0 0 7

D 0 1

D 0 2

Endoplasmin OS=Homo sapiens GN=HSP90B1 PE=1 SV=1 [ENPL_HUMAN] 8 8 1 5 T-complex protein 1 subunit beta OS=Homo sapiens GN=CCT2 PE=1 SV=4 - [TCPB_HUMAN]

Stress-70 protein, mitochondrial OS=Homo sapiens GN=HSPA9 PE=1 SV=2 - [GRP75_HUMAN]

Proteolysis Acylamino-acid-releasing enzyme OS=Homo sapiens GN=APEH

G 0 6

7 6 2 6

G 0 1

P 7 8 3 7 1 P 3 8 6 4 6

6 7

B 0

P 1

2 4 0 9

E 0 2

S P R M E D 4

H S P D 1 H S P 9 0 B 1 C C T 2

H S P A 9

A P

0 4 3 1 0 . 0 3 7 3

2. 49 4

2. 23 4

2. 77 3

3. 03 9

3. 88 1

2. 57 2

14

1.3 19

20. 4/4

1.1 59

19. 2/5

1.4 71

11. 2/5

1.6 04

50. 6/3 6 0 . 0 2 2 5

0 . 0 1 3 8 0 . 0 5 6 2

85 6

3. 32 8

1.7 34

29. 1/2 8

21. 5/1 0

1.9 56

1.3 63

41. 2/3 2 0 .

10

3.4

7.9 /6

1/ 7. 3

6. 8

. 3 8

2 9. 7/ 5. 1

33 .4/ 4. 9

1 3 . 1 9

3 4. 3/ 6. 0

40 .4/ 6. 4

4 0 . 3 6

6 1. 0/ 5. 9

61 .0/ 4. 9

2 4 . 2 6

6 1. 0/ 5. 9

59 .4/ 5. 1

9 2. 4/ 4. 8

93 .1/ 4. 9

6 0 9 . 2 4 1 5 5 . 6 3

5 7. 5/ 6. 5

58 .7/ 7. 4

3 6 . 6 1

7 3. 6/ 6. 2

69 .5/ 5. 2

2 3 8 . 2 7

8 1.

75 .5/

2 3

PE=1 SV=4 - [ACPH_HUMAN]

Acylamino-acid-releasing enzyme OS=Homo sapiens GN=APEH PE=1 SV=4 - [ACPH_HUMAN]


Protein modification Protein phosphatase 1 regulatory subunit 7 OS=Homo sapiens GN=PPP1R7 PE=1 SV=1 [PP1R7_HUMAN]

Ubiquitin-like modifier-activating enzyme 1 OS=Homo sapiens GN=UBA1 PE=1 SV=3 [UBA1_HUMAN]

Protein synthesis Elongation factor 1-delta OS=Homo sapiens GN=EEF1D PE=1 SV=5 - [EF1D_HUMAN]

Protein disulfide-isomerase OS=Homo sapiens GN=P4HB PE=1 SV=3 - [PDIA1_HUMAN]

Procollagen-lysine,2-oxoglutarate 5-dioxygenase 3 OS=Homo sapiens GN=PLOD3 PE=1 SV=1 - [PLOD3_HUMAN]

Cell morphology and transport Cytoskeleton Beta-actin-like protein 2 OS=Homo sapiens GN=ACTBL2 PE=1 SV=2 - [ACTBL_HUMAN]

Src substrate cortactin OS=Homo sapiens GN=CTTN PE=1 SV=2 [SRC8_HUMAN]

1 9

7 7 0 3

9 3 1 5

6 9 3 0

8 2 1 1

8 6 2 0

5 7 0 8

7 3 2 2 7 7 3

3

3 7 9 8 P 1 3 7 9 8

E H

Q 1 5 4 3 5 P 2 2 3 1 4

P P P 1 R 7 U B A 1

E E F 1 D

F 1 2

P 2 9 6 9 2 P 0 7 2 3 7 O 6 0 5 6 8

G 0 3 A 0 2

Q 5 6 2 R 1 Q 1 4

A C T B L 2 C T T

A 0 1

B 0 4

A 0 9

F 0 9

C 0 4

0 2 0 8 0 . 0 0 6 0

A P E H

P 4 H B

P L O D 3

0 . 0 4 4 5 0 . 0 0 7 4 0 . 0 1 8 2 0 . 0 4 0 5 0 . 0 0 6 4

0 . 0 0 0 9 0 . 0

3. 43 5

2. 00 7

49 .1 07 28 .5

.7 00

78 .4 09

22 .9 03

8. 36 5

1. 85 7

20

6.2 93

4.5 17

3.0 64

0.8 93

1.7 80

5.3 /2

7.6 /6

35. 6/1 0

39. 6/2 2

22. 4/1 4

1.0 05

5.6 18 4.8 37

6.4 /5

0 . 0 0 6 6

23 .8 79

4.5 78

16. 0/5 18. 4/8

2/ 5. 5

5. 4

. 7 7

8 1. 2/ 5. 5

76 .7/ 5. 3

1 8 . 7 5

4 1. 5/ 4. 9

44 .3/ 4. 9

8 . 6 6

1 1 7. 8/ 5. 8

10 9. 3/ 5. 4

2 0 . 0 9

3 1. 1/ 5. 0

34 .6/ 4. 9

5 8 . 7 2

5 7. 1/ 4. 9

60 .4/ 4. 9

8 4. 7/ 6. 0

76 .6/ 5. 8

1 2 7 . 2 7 9 6 . 4 2

4 2. 0/ 5. 6

41 .6/ 5. 1

4 9 . 0 9

6 1. 5/

68 .2/ 5.

2 6 .

4

Tropomodulin-3 OS=Homo sapiens GN=TMOD3 PE=1 SV=1 - [TMOD3_HUMAN]


Villin-1 OS=Homo sapiens GN=VIL1 PE=1 SV=4 [VILI_HUMAN]



8 2 0 2

4 7 1 3

4 7 1 0

4 7 0 7

C 0 9

E 0 9

E 1 0

E 1 1

Junction plakoglobin OS=Homo sapiens GN=JUP PE=1 SV=3 [PLAK_HUMAN] 5 1 2 Cell-Cell junction Cingulin OS=Homo sapiens GN=CGN PE=1 SV=2 [CING_HUMAN]

Desmoglein-1 OS=Homo sapiens GN=DSG1 PE=1 SV=2 [DSG1_HUMAN]

Catenin delta-1 OS=Homo sapiens GN=CTNND1 PE=1 SV=1 - [CTND1_HUMAN]

Cell polarisation Shootin-1 OS=Homo sapiens

A 0 6

2 4 7 Q 9 N Y L 9 P 0 9 3 2 7 P 0 9 3 2 7 P 0 9 3 2 7 P 1 4 9 2 3

N

2 1 5

85

T M O D 3 V I L -1

V I L -1

V I L -1

0 . 0 2 6 4 0 . 0 3 3 6 0 . 0 4 1 8 0 . 0 3 5 2

4. 18 6

8. 37 4

3. 79 0

1.2 45

12. 8/4

19. 7/1 5

2.0 66

20. 0/1 4

3.0 66

25. 4/1 9

1.9 22

J U P

0 . 0 1 2 7

65 .9 45

6.0 43

9.9 /7

2

3 0

3 9. 6/ 5. 2

39 .7/ 5. 0

1 8 . 0 4

9 2. 6/ 6. 4

89 .6/ 6. 3

6 3 . 7 6

9 2. 6/ 6. 4

89 .5/ 6. 1

5 8 . 6 1

9 2. 6/ 6. 4

89 .8/ 6. 2

9 7 . 4 7

8 1. 7/ 6. 1

64 .2/ 5. 8

2 3 . 6 2

10 4. 4/ 5. 4

1 3 . 2 9

10 3. 5/ 5. 4

1 6 . 4 0

10 1. 8/ 6. 8

1 1 . 3 5

74

3

C G N

3 7 0 7

Q 9 P C 2 0 M 3 7 Q 0 2 B 4 0 1 2 3 O 6 0 A 7 0 1 7 6

C T N N D 1

0 . 0 1 3 7

26 .1 73

4.7 10

6.9 /6

1 3 6. 3/ 5. 5 1 1 3. 7/ 5. 0 1 0 8. 1/ 6. 2

5

C

A

K

0

-

-

12.

7

6 9 1 0

6 8 1 6

0 . 0 0 7 0

2. 37 0

5. 4

13 .0 10

3.7 02

D S G 1

5.8 5/6 0 . 0 2 2 1

13 .0 39

3.7 05

6.0 /5

GN=KIAA1598 PE=1 SV=4 [SHOT1_HUMAN]


Cell spreading Rho guanine nucleotide exchange factor 7 OS=Homo sapiens GN=ARHGEF7 PE=1 SV=2 [ARHG7_HUMAN]

Endocytosis Endophilin-A2 OS=Homo sapiens GN=SH3GL1 PE=1 SV=1 [SH3G1_HUMAN]

Endosome Early endosome antigen 1 OS=Homo sapiens GN=EEA1 PE=1 SV=2 - [EEA1_HUMAN]

Protein Hook homolog 2 OS=Homo sapiens GN=HOOK2 PE=1 SV=3 [HOOK2_HUMAN]

Organelle transport Kinesin light chain 4 OS=Homo sapiens GN=KLC4 PE=1 SV=3 [KLC4_HUMAN]

Signal transduction Rab GDP dissociation inhibitor beta OS=Homo sapiens GN=GDI2 PE=1 SV=2 [GDIB_HUMAN]

Ras GTPase-activating proteinbinding protein 1 OS=Homo sapiens GN=G3BP1 PE=1 SV=1 [G3BP1_HUMAN]

6 0 1

3 8 3 7

7 3 1 7

7 9 1 2

6 7 1 2

5 6 2 1

5 3 1 5 6 5 1 8

0 5

0 M Z 6 6

I A A 1 5 9 8

Q 1 4 1 5 5

A R H G E F 7

0 . 0 4 2 7

Q 9 9 9 6 1

S H 3 G L 1

0 . 0 0 6 2

E A A 1

F 0 7

Q 1 5 0 7 5 Q 9 6 E D 9

0 . 0 2 5 0 0 . 0 1 1 6

E 0 5

Q 9 N S K 0

K L C 4

P 5 0 3 9 5 Q 1 3 2

G D I2

E 0 1

B 0 8

A 1 0

G 0 2 C 0 1

H O O K 2

G 3 B P

. 0 1 5 6

0 . 0 2 3 6 0 . 0 0 8 7 0 . 0 0

3. 35 2

3. 91 7

13 .0 10

2. 14 6

94 .4 20

32 .2 58

1. 98 7 38 .0 48

1.7 45

0/7

1.9 70

29. 9/1 0

1.1 02

21. 55/ 27 0 . 0 1 2 9

17 .4 65

4.1 26

5.0 12

7.1 /5

4.2 /2

30. 1/1 4

0.9 91 5.2 50

.5/ 5. 9

5 . 8 3

9 0. 0/ 7. 1

91 .7/ 6. 6

1 6 . 2 4

4 1. 5/ 5. 4

40 .4/ 5. 3

3 9 . 5 8

1 6 2. 4/ 5. 7 8 3. 2/ 5. 5

11 1. 5/ 5. 3

1 1 2 . 9 4 1 6 . 3 3

7.5 /6

3.7 02

6.5 61

1. 6/ 5. 3

0 . 0 0

43 .2 03

5.4 33

24. 9/9

78 .7/ 5. 3

6 8. 6/ 6. 2

64 .9/ 5. 9

4 . 1 7

5 0. 6/ 6. 5

46 .2/ 5. 7

8 2 . 6 6

5 2. 1/ 5.

59 .9/ 5. 4

7 2 . 7

Nodal signaling Nodal modulator 1 OS=Homo sapiens GN=NOMO1 PE=1 SV=5 - [NOMO1_HUMAN]


Other functions Calcium binding Annexin A2 OS=Homo sapiens GN=ANXA2 PE=1 SV=2 [ANXA2_HUMAN]

Cell cycle and proliferation Nuclear migration protein nudC OS=Homo sapiens GN=NUDC PE=1 SV=1 - [NUDC_HUMAN]

DNA replication licensing factor MCM7 OS=Homo sapiens GN=MCM7 PE=1 SV=4 [MCM7_HUMAN]

14-3-3 protein gamma OS=Homo sapiens GN=YWHAG PE=1 SV=2 - [1433G_HUMAN]

6 9 0 6

3 2 1 5

8 2 1 6

4 7 1 2

9 1 0 5

8 3

1

8 2

8 1

5

C 0 8

Q 1 5 1 5 5

N O M O 1

0 . 0 0 6 8

0 . 0 1 2 0

1 3 4. 2/ 5. 8

11 1. 1/ 5. 5

5 3 . 7 6

C 0 6

P 0 7 3 5 5

A N X 2

3 8. 6/ 7. 7

33 .7/ 6. 9

7 . 8 2

Q 9 Y 2 6 6 P 3 3 9 9 3 P 6 1 9 8 1

N U D C

0 . 0 2 2 6 M 0 C . M 0 7 1 5 7 Y 0 W . H 0 A 0 G 6 8

3 8. 2/ 5. 4

33 .6/ 4. 9

3 0 . 6 0

8 1. 3/ 6. 5

77 .9/ 6. 0

4 2 . 2 6

2 8. 3/ 4. 9

27 .5/ 5. 1

7 5 . 5 9

O 0 0 3 3 8

S U L T 1 C 2

3 4. 9/ 7. 5

38 .4/ 5. 1

7 . 5 7

C 0 7

B 0 9

B 0 7

Carcinogenesis Sulfotransferase 1C2 OS=Homo sapiens GN=SULT1C2 PE=1 SV=1 - [ST1C2_HUMAN] 7 7 2 8

F 0 2

4. 32 7

2.1 14

0 . 0 1 5 3

0 . 0 0 5 2

23 .7 41

4. 51 7

2. 79 4

3. 60 4

4.5 69

2.1 75

1.4 82

1.8 50

3. 77 0

3. 20 6

1.9 14

1.6 81

15. 8/1 5

10. 6/3

18. 7/6

14. 6/8

24. 7/8

7.8 /2

8

Table 2. List of de-regulated tentatively annotated metabolites in the cell extract. Metabolites have been assigned HMBD tags and have been grouped according to their biological functions. Quantitative changes after AuNPs treatment (5 nm AuNPs vs Ctrl and 30 nm AuNPs vs Ctrl) are reported (log2.fold change) for individual metabolites. Log 2 = 0.58, thus the range was set from -0.58 to 0.58 and colour-coded. Green for < -0.58, 0 to black and > 0.58 to red. The values in between are shown as colour gradients.


Table 2

HMDB tags

5 nm AuNPs log2.Fo ld change

30 nm AuNPs log2.Fo ld change

1.0663

0.0978

1.9865

0.0989

1.9865

0.0989

0.3111

0.0663

0.3111

0.0663

0.2819

0.3951

-0.0346

0.1305

-0.0346

0.1305

-0.0346

0.1305

-0.0346

0.1305

-0.0346

0.1305

-0.0882

-0.0244

-0.0346

0.1305

-0.0346

0.1305

m/z

rt (s)

76.03957 7 194.0823 49 194.0823 49 218.1401 62 218.1401 62 308.0932 31 132.0774 53 132.0774 53 132.0774 53 132.0774 53 132.0774 53

606.7 4 606.7 8 606.7 8 474.9 6 474.9 6 465.7 8 282.9 8 282.9 8 282.9 8 282.9 8 282.9 8

188.0717 74 132.0774 53 132.0774 53

548.3 4 282.9 8 282.9 8

Metabolism Amino Acids glycine (R)-1,2-dimethyl-5,6-dihydroxytetrahydroisoquinoline 5,6-Dihydroxyindole-2-carboxylic acid N-a-Acetylcitrulline Deoxyhypusine Glutathione L-Glutamic gamma-semialdehyde L-Isoleucine L-Leucine Iminoaspartic acid 4-Hydroxy-L-proline

HMDB001 23 HMDB124 84 HMDB012 53 HMDB008 56 HMDB111 50 HMDB001 25 HMDB021 04 HMDB001 72 HMDB006 87 HMDB011 31 HMDB060 55

amino acids catabolism 2-Keto-6-acetamidocaproate Propionylglycine 5-Amino-2-oxopentanoic acid Carbohydrate metabolism

HMDB121 50 HMDB007 83 HMDB062 72

Propionylcarnitine

HMDB008 24

218.1401 62

474.9 6

114.0918 82 114.0918 82

579.1 5 579.1 5

0.3951

308.0932 31

465.7 8

1.9865

0.0989

194.0823 49

606.7 8

-0.0346

0.1305

1.0663

0.0978

132.0774 53 76.03957 7

282.9 8 606.7 4

0.2399

0.2129

0.2399

0.2129

0.2399

0.2129

261.1461 24 261.1461 24 261.1461 24

559.7 3 559.7 3 559.7 3

-0.0346

0.1305

-0.0346

0.1305

-0.0346

0.1305

132.0774 53 132.0774 53 132.0774 53

282.9 8 282.9 8 282.9 8

1.0663

0.0978

0.2444

0.2150

76.03957 7 220.1193 91

606.7 4 533.4 1

1.9865

0.0989

1.9865

0.0989

194.0823 49 194.0823 49

606.7 8 606.7 8

HMDB122 04

0.2444

0.2150

220.1193 91

533.4 1

HMDB021 14

0.6721

-0.0079

309.1287 99

640.6 5

0.3111

0.0663

-0.0335

0.9237

-0.0335

0.9237

HMDB011 77

0.2819

HMDB008 59

Glutamate 1-Pyrroline-5-carboxylic acid 1-Pyrroline-2-carboxylic acid

HMDB013 01 HMDB068 75

Glutamate catabolism (S)-Succinyldihydrolipoamide Fatty acids Methylhippuric acid


Energy E-transport Beta-Guanidinopropionic acid Trimethylamine N-oxide

HMDB132 22 HMDB009 25

Protein associated Products of Protein degradation L-gamma-glutamyl-l-isoleucine L-gamma-glutamyl-l-leucine L-alpha-glutamyl-l-hydroxyproline

HMDB111 70 HMDB111 71 HMDB111 61

other 4-Hydroxyproline N-Acetyl-L-alanine 3-Hydroxy-L-proline

HMDB007 25 HMDB007 66 HMDB021 13

Secondary metabolism Vitamines 1-Amino-2-propanol Pantothenic acid

HMDB121 36 HMDB002 10

Melanin biosynthesis L-dopachrome D-dopachrome

HMDB014 30 HMDB116 22

Plant hormone cis-Zeatin other Bisdemethoxycurcumin

Metabolites with unclear classification Gamma-glutamyl-L-putrescine dCMP

HMDB122 30 HMDB012 02

0.3111

0.0663

0.2819

0.3951

218.1401 62 308.0932 31

474.9 6 465.7 8

80.94817 9 82.94523 5 96.92215 1 138.9070 97 140.9041 52 331.1105 36 353.1189 00

301.1 9 300.9 5 300.2 4 301.3 2 301.4 6 640.4 8 683.3 0


Unidentified Metabolites unidentified

unknown

0.4930

-0.0967

unidentified

unknown

0.5106

-0.1012

unidentified

unknown

0.3434

0.1163

unidentified

unknown

0.5903

-0.1402

unidentified

unknown

0.5683

-0.1239

unidentified

unknown

0.7859

-0.1193

unidentified

unknown

-0.3976

-0.0550

Table 3. Identified molecular networks using Ingenuity IPA. The table reports the most significant molecular networks in response to 5 and 30 nm AuNPs treatments, by analyzing the differentially expressed proteins (listed in Table 1) and metabolites (listed in Table 2) from the cells. The network number, the list of all proteins and metabolites involved in the network, the number of molecules overlapping between our data set and the network and the top functions related to the network are shown. Focus molecules are indicated in bold and de-


regulation is indicated with a colored arrow (red: up-regulated, green: down-regulated).

Figure captions Figure 1. Experimental design. A combination of 2D-gel based proteomic and MS-based metabolomic approaches was used to analyse the differentially expressed proteome and metabolites of the cytoplasmic compartment of Caco-2 cells exposed to 5 or 30 nm AuNPs (300μM) for 72 h. Data obtained were interpreted using a combination of bioinformatics tools


for a combined omics approach.

Figure 2. Proteomic analysis of the cytoplasmic extract of Caco-2 cells exposed to AuNPs. Representative two-dimensional gel protein maps of cytoplasmic fractions of (A) untreated (Ctrl), (B) treated with 5 nm AuNPs, and (C) treated with 30 nm AuNPs cells for 72 h. (D) The Venn diagram is showing the distribution of differentially expressed proteins: 5nm AuNPs vs Ctrl (red circle), 30 nm AuNPs vs (blue circle) or 5 vs 30 nm AuNPs (green circle). Number inside overlapping region of two circles refers to the spots common to


different groups.

Figure 3. Biological functions altered by AuNPs exposure. Bar charts grouped by biological function representing the differentially expressed proteins and metabolites, after 72 hr of treatment with AuNPs. (A) Represents the de-regulated proteins by 5 nm or 30 nm AuNPs (B) the de-regulated metabolites following exposure to 5 and 30 nm AuNPs.

A

Biological functions of differentially expressed proteins

B

14 5 nm AuNPs

12

30 nm AuNPs Number of proteins


10 8 6 4 2 0

Metabolism

Energy

Transcription

Protein

Cell morphology Signal Other functions & transport transduction Biological functions

Figure 4. Molecular networks. De-regulated molecular networks in response to 5 or 30 nm AuNPs exposure in Caco-2 cells (300 µM, 72 hr). The networks are obtained by analysing the differentially expressed proteins and metabolites (listed in Tables 2 and 3) using Ingenuity IPA. Identified de-regulated proteins and metabolites involved in the network are highlighted in bold. The colour indicate the de-regulation (red: up-regulated, green: downregulated). A: Cellular compromise (degeneration); B: Small Molecule Biochemistry; C: Cell Morphology; D: Cellular Assembly and Organization; E: Cellular Assembly and Organization according to Table 6.


Supplementary Methods Gold nanoparticles synthesis Gold (III) chloride trihydrate (> 99.9%, HAuCl4 3H2O), trisodium citrate dihydrate (> 99.9%, Na-Citrate) and sodium borohydride (99.9%, NaBH4), sodium hydroxide (NaOH) were purchased from Sigma-Aldrich and used as received without further purification. Before use, AuNPs were concentrated by Centrifugal filter units (Amicon centrifugal ultrafiltration unit,


Millipore), which were washed three times with milliQ-water at 3500 rcf The synthesis of nominal 5 nm AuNPs 5 nm was performed as described in the literature (Coradeghini et al., 2013). The synthesis of AuNPs stabilised with Na-Citrate (2.5 mM of NaCitrate at pH 6.7), was carried out by reduction of HAuCl4 3H2O salt with the strong reducing agent NaBH4. Briefly, 95 mL of MilliQ water were stirred in an ice bath for 2 h. Then 5 mL of aqueous solution of HAuCl4 3H2O (10 mM) and 2.5 mL of Na-Citrate (100 mM) were added to the water solution during the stirring. Afterwards, 1 mL of aqueous NaBH4 (100 mM) was added to the solution under vigorous stirring. The reduction of gold salt to AuNPs by NaBH4 produced a change in colour of the solution from pale yellow to dark red. The suspension was then stirred in the ice bath for further 10 min and then left to warm to room temperature. The nominal concentration of Au calculated by the stoichiometry was 0.5 mM. The synthesis of nominal 30 nm AuNPs was carried out by a two-step seed growth method in which the reduction of HAuCl4 3H2O salt in the presence of Na-Citrate was used to selectively enlarge highly mono-dispersed 12 nm size gold seed particles. In the first step of the process, the 12nm seed particles were obtained by an adapted previously available method (John Turkevich, 1951). Briefly, the solution was heated up using a microwave apparatus (Discover S by CEM corporation) to ensure a highly reproducible rapid heating. In this method, 5 mL of 10 mM HAuCl4 3H2O was dissolved in 95 mL of water. The solution

was rapidly heated up and held at 97 ºC for 5 min using a microwave power of 250 W under vigorous mechanical stirring; 2.5 ml of Na-Citrate (100 mM) was added to the solution and kept at 97 ºC for another 20 min. Afterwards, the solution was rapidly cooled down to 40 ºC in a flow at compressed nitrogen and then to room temperature. AuNPs were produced by the reduction of the gold salt by Na-Citrate that acted as both reducing agent and stabilizer. The final solution contained 12 nm size gold nanoparticles with 0.5 mM of gold NPs stabilized


with 2.5 mM of sodium citrate at pH 6.7. In the second stage of the synthesis, 95 mL of MilliQ water were left to stir at 60 ºC for two h; 2.8 mL of Na-Citrate (100 mM) and 0.420 mL of NaOH 200 mM were added to the NaCitrate solution and left to equilibrate for 30 min and, then, 2.24 mL of 10 mM of HAuCl 4 3H2O) and 0.6 mL of 12 nm gold nanoparticles with a gold NPs concentration of 0.5 mM were added to the solution, under vigorous stirring. The solution was left to react for 48 h at 60 ºC. The final solution of Au was 0.23 mM of gold NPs stabilised with 2.8 mM of NaCitrate at pH 6.7. After synthesis, 5 and 30 nm AuNPs were characterized and then concentrated by centrifugal ultrafiltration using Amicon filter of 10 kDa MWCO (2000 rcf, 10 min). The final concentration of Au was determined by UV-vis Spectroscopy (Liu et al., 2007). The final concentration for 5 and 30 nm AuNPs was 4.24 mM and 4.06 mM respectively.

Gold nanoparticles characterization Particle size distributions of the AuNPs was determined by Centrifugal Liquid Sedimentation (CLS) measurements on the as-synthesized nanoparticle dispersions. Measurements were carried out on CLS instrument model DC24000UHR by CPS Instruments Inc. Measurements were made in an 8 wt%-24 wt% sucrose density gradient with a disc speed of 22000 rpm.

Before each sample injection of 100 µL, a calibration step was performed using certified PVC particle size standards with weight mean size of 380 nm. Moreover, AuNPs were visualized using a Transmission Electron Microscope TEM (JEOL 2100, Japan) at an accelerating voltage of 200 kV. 5 µL of AuNPs solution were spotted onto ultrathin Formvarcoated 200-mesh cupper grids (Tedpella Inc.) and left to dry in air at 4ºC. TEM images were analysed with ImageJ to obtain the average size and the size distribution for each sample. The particle size distribution was also determined by using Malvern Zetasizer Nano-ZS


instrument. Z-potential was measured by Zetasizer Nano ZS instrument (Malvern Instruments, UK) and recorded in a DTS1060C disposable cell with an equilibration time of 120 sec. Measurement were performed just after measuring the pH of NPs solution. Size distribution data of the AuNPs after synthesis using CLS and DLS are reported in Supplementary Table 1. TEM pictures were collected to verify the spherical shape of AuNPs as shown in Figure 6S. The Z-potential of the samples was also calculated.

2D gel-based proteomic experiments 2D electrophoresis First-dimension isoelectric focusing (IEF) was carried out on a Multiphor II system (GE Healthcare, Little Chalfont, UK). Pre-cast immobilised pH gradient strips (GE Healthcare) (non-linear pH range 3.0 – 10.0) were rehydrated for 18 h. The protein samples (120 μg) were loaded at the anode and IEF was run for 110 kVh at 18◦C (Gioria et al., 2014). After isoelectric focusing, the strip gels were equilibrated with a solution containing 50 mM Tris-HCl (pH 8.3), 6 M urea, 4% SDS, 30% glycerol and 1% dithiothreitol (Sigma-Aldrich®) for 15 min twice and once again in the same buffer containing 2.6% iodoacetamide instead of dithiothreitol and 0.4% bromophenol blue (BioRad, Italy). Second dimension separation

(SDS-PAGE) was carried out with a Hoefer Standard Dual Cooled Vertical Unit. The equilibrated gel strips were embedded at the top of a stacking gel in 1% molten agarose containing 0.8% bromophenol blue and separated in the second dimension according to their molecular mass using 8 – 14% linear gradient sodium SDS-PAGE at 40 mA for 5 h. After electrophoresis the gels were stained with fluorescent dye Sypro Ruby (Molecular Probe Inc.,


Lifetechnologies, Italy) for protein pattern analysis.

Differential image analysis of 2D protein patterns Block randomized design on six biological replicates for each experimental condition (Control, 5 or 30 nm AuNPs treatments) was performed to reduce the bias and variance in the 2D-gel protein patterns in order to better estimate the difference between untreated and AuNPs treated cells. After the electrophoretic separation, Sybro Ruby-stained 2D gels were scanned with a GS800 imaging densitometer (BioRad) under the same scanning conditions. Background subtraction, spot detection, gel alignment and spot matching were performed using PDQuest v. 7.3.0 software package (BioRad). To compensate for subtle differences in sample loading and gel staining, the volume of each spot was normalized according to the total number of valid spots in each gel. The 2D gels were sorted in three classes (Control, 5 or 30 nm AuNPs) and differential analysis was performed on logarithmic transformed data to improve the variance across all spots in a gel. A statistical data analysis (Mann-Whitney test) was performed to identify differentially regulated proteins in the whole experiment in pairwise comparisons of 5 nm AuNPs or 30 nm AuNPs treatment vs control (Control) for each of the 6 biological replicates. Values of the

spot volume ratio falling outside the range of ± two folds change were regarded as being significantly different between control and treated samples. Apparent molecular weight (MW) and isoelectric points (pI) were established bycomparison with known proteins used as internal standards.

Preparative 2D gels and protein spot picking


A preparative experiment was run using 200 μg of protein from control and 5 nm AuNPs treated samples. Experimental conditions for electrophoresis were the same as the ones described for the analytical gel. The gels were Sypro Ruby stained and digitized for image analysis. Preparative gels were matched with analytical gels for protein selection in the 2D map using PDQuest software. Corresponding spots were listed and numbered accordingly for further MS/MS identification. Selected protein spots were excised and transferred to a 96well plate using a ProteomeWorks Plus Spot Cutter System (BioRad).

In-gel protein hydrolysis and peptide extraction The experimental conditions were described in details in our previous work (Gioria et al., 2014).

Protein identification by MS LC-MS/MS analyses were performed on a Ultimate 3000 liquid chromatography system (Dionex, Sunnyvale, CA, USA) coupled on-line with a LTQ Orbitrap XL Fourier Transform mass spectrometer (Thermo Scientific™, Waltham, MA, USA) equipped with a nanoelectrospray ionisation source (nano ESI).

The experimental conditions of the binary LC system were described in detail in our previous work (Gioria et al., 2014). The LTQ Orbitrap instrument was operated with Xcalibur™ software (Thermo Scientific™) in the data dependent analysis mode, switching between MS and MS/MS analysis. This particular experimental mode involves two separate scan events (events 1 and 2). In parallel to the full-scan MS acquisition in the Orbitrap (event 1), the MS/MS fragmentation of the 6 most abundant precursor ions was simultaneously achieved in the ion trap (event 2). For event 1, a mass range of 350 – 2,000 was selected for the Orbitrap,


a resolution of 60,000 was set (average scan time of 1 s) and data were acquired in the continuum mode. For event 2, a mass range of 50 – 2,000 was selected with the ion trap and data were acquired in the centroid mode. Fragmentation scans were done using collision induced dissociation (CID), allowing masses of both the parent peptide and its daughter ions to be detected. The accurately measured masses of the tryptic peptide and its fragments were further used to search for protein candidates in protein sequence databases. Proteomic data processing The fragment ion spectra obtained in the data dependent analysis mode were processed using XCalibur software v. 2.1 (Thermo Scientific™), a software that converts MS/MS raw data to peak lists. After centroiding and background subtraction, the peptide mass lists were searched with

Proteome

Discoverer

v1.3

(Thermo

Scientific™,

Italy)

against

the

UniProt_Human_v3.87 database (http://www.uniprot.org/). The SEQUEST workflow was used and proteins were identified by correlating the processed mass spectral data to protein entries. The main search parameters were: enzyme, trypsin; no restriction on molecular weight and pI; taxonomy, Homo sapiens; two missed cleavage allowed; tolerance on the precursor mass: 10 ppm; tolerance on fragment mass, 0.8 Da; fixed modification: carbamidomethylation of cysteine; variable modification: oxidation of methionine; other potential post-translational modifications could also be considered.

After databank searching and sequence analysis each identified peptide was assigned a peptide score and a probability based score algorithm, which gave an indication for the reliability of the peptide identification. Proteins listed as significant matches in the Proteome Discoverer search results were considered as good candidates for identification if the following conditions were met: (a) sequence coverage of at least 5%, (b) fair agreement of theoretical MW and pI with the respective experimental values obtained from 2D gel image analysis (differences allowed: 25% for MW and 10% for pI), (c) probability score of at least


10. If more than one protein met the above criteria for a given spot, the protein with the highest number of matched peptides was retained.

MS-based metabolomic experiments Metabolomic analysis For metabolomic experiments the LTQ Orbitrap XL mass spectrometer was equipped with an electrospray ionisation source (ESI). Sample were loaded and pre-concentrated on a Waters® Atlantis T3 guard cartridge (2.1 x 10 mm, 3 m) and metabolites were separated on an Atlantis T3 column (1.0 x 150 mm, 3 m). The micro LC pump was operated at a flow rate of 40 μl/min obtained after a 1:4 micro flow split (Dionex). The eluents used were (A) 0.1% HCOOH in Milli-Q H2O, (B) 0.1% HCOOH in CH3CN and (C) 0.1% HCOOH in CH3OH. The linear gradient used to achieve the analytes separation was as follows: 0 – 1.5 min: 95% A, 5% C; 10 min: 2% A, 48% B, 50% C; 10 – 11 min: 2% A, 48% B, 50% C; 11.1 min: 2% A, 98% B; 11.1 – 12 min: 2% A, 98% B; 12 min: 2% A, 98% B; 12.1 min: 98% A, 2% C; 12.1 – 17 min: 98% A, 2% C. The LTQ Orbitrap instrument was operated in the MS mode using the FTMS (Orbitrap) mass analyser. Optimised settings were as follows: capillary voltage, 45 V; Tube lens voltage,

135 V; spray voltage, + 3.5 kV; transfer capillary temperature, 200 °C. The Orbitrap analyser was calibrated using a solution of caffeine, MRFA (L-methionyl-arginyl-phenylalanylalanine acetate x H2O) and Ultramark 1621 (all from Sigma – Adrich®, Italy) in the mass to charge (m/z) range of 50 – 1,000. The MS acquisition was performed over the full m/z range with a scan time of 0.7 s. Resolution used was 30,000 FWHM and mass accuracy was in the range 1 – 5 ppm. LC-MS data were generated in continuum mode.


Experiment design for metabolomics involves the analysis of QCs (to establish the repeatability and intermediate precision of the LC-MS method), pool of samples (for tracking intra-batch and inter-batch differences), analytical blanks (for possible contamination) and the study samples. Control and treated samples were run in randomized order with the analytical blanks, QCs and pools analyzed 9 times during the sequence.

Metabolomics data processing An experimental design table was created for each of the 6 batches of analysis. To each batch was associated a .csv file containing the following information on the analysis sequence: sample name (pool, standard, blank, sample), sample code (unique code according to the standardised sample label format), file name (unique name in a sequence), treatment, nanoparticle size (5 or 30 AuNPs), cell model (Caco-2 cells) and sample type. Raw data files were converted into Network CommonData Form (NetCDF) format using the "File Converter" tool of XCalibur software and placed on a server ready for data processing. LC-HR-MS data processing was performed using the open-source XCMS software package (http://masspec.scripps.edu/xcms/xcms.php) (version 1.14.1 running under R version 2.8.1) that uses several algorithms written in R language (Smith et al., 2006, Kuhl et al., 2012). A Rpackage, called "MCG" was developed in-house.

Data pre-processing The first step was to filter and detect the peaks. For this, the ion chromatograms were extracted and the "matched filter" algorithm applied. The peak detection algorithm is based on splitting the LC-MS data, a fraction of a mass unit wide, and then operating on those individual parts in the chromatographic time domain. Extraction was performed as follows: rtcut = c(2,16)*60, profmethod = "bin", fwhm = 30, max = 10, snthresh = 10, step = 0.01,


steps = 2, mzdiff = 0.013. After detecting peaks in individual samples, the peaks were matched across samples to allow calculation of retention time deviations and relative ion intensity comparison. This is accomplished using the density algorithm (for chromatographic data). Optimised settings were: minsamp = 2, minfrac = 0.5, bw = 5, minfrac = 0.5 minsamp = 2, mzwid = 0.013, mzVsRTbalance = 10, mzCheck = 0.2, rtCheck = 15, kNN = 10. For the function Groupval, parameters were as follows: DiffClass = c("ANOVANw"), UseDiff = c(p = 0.05, fc = 2, camera = 0.05)). These groups are then used to identify and correct drifts in retention time from run to run. Chromatograms were aligned using the "Obiwarp" algorithm (Prince and Marcotte, 2006). For every group, the median retention time and the deviation from median for every sample in that group was calculated. Missing peaks can occur because they were not observed during the peak detection step or because the analyte is not present in the sample. Therefore, an additional step called 'fill peak' was used. To account for the missing peak data, a mathematical function was applied to approximate the difference between deviation and interpolation in sections where no peak groups are present.

Normalisation was applied to correct for changes in the sensitivity of the detector across the analytical sequence or changes in the sample concentration. In the case of cell culture medium samples, the data matrix obtained from XCMS was normalised to a common MS total useful signal (MSTUS): i.e. the sum of the areas of the peaks that are found in all the chromatographic runs in the set. It is based on the calculation of a most probable dilution factor by looking at the distribution of the quotients of the amplitudes of a test spectrum by


those of a reference spectrum. A peak intensity table with m/z, retention time (rt) and intensities was generated for each of the 6 batches of analysis and submitted for statistical treatment.

Statistical data analysis A statistical data analysis was performed to identify differentially regulated metabolites in pairwise comparisons of 5 and 30 nm AuNPs treatment vs control for each of the 6 biological replicates. A Welch's t-test (independent two-sample t-test, unequal variances) was used for comparison of treated cells against untreated cells. The program calculated a fold change and carried out the test between samples from two nominated groups (5 or 30 AuNPs) and ranked the m/z and rt values in order of the t-statistic (or p). The features that failed to have an adjusted p < 0.05 for the test were rejected. A fold-change (FC) was calculated within experiment groups (5 or 30 AuNPs) for each feature. The FCs were computed from the average values across biological replicates. This method assessed formally whether the true differential expression is greater than a predefined FC criteria (threshold = 2). Metabolites were considered to be differentially expressed if they showed a fold-change of at least 2 (up- or down-regulated).

Further peak redundancy removal (de-isotoping and de-adduction) was carried out by autocorrelating intensity feature values across quality control samples into the pc group windows calculated by the CAMERA open source package (Kuhl et al., 2012), identifying adducts and isotopes occurring from the parent molecular ions. Data from the 6 independent analysis batches (corresponding to the 6 biological replicates) were combined for processing. The m/z features retained after statistical analysis were then


submitted to database search for annotations.

Mass-based metabolite identification Confidence in measurement and reporting standards for metabolite identifications are essential for integrating metabolomics data with data from other omics disciplines, such as proteomics.

In

2007,

the

Metabolomics

Standards

Initiative

(MSI)

(http://msi-

workgroups.sourceforge.net) of the Metabolomics Society proposed reporting standards in metabolomics (Sumner et al., 2007). These standards recommend that authors should report the level of identification for all metabolites base on a four-level system ranging from level 1 (identified compounds), via levels 2 and 3 (putatively-annotated compounds) and 3 (putatively-compound classes) to level 4 (unidentified or unclassified metabolites which nevertheless can be differentiated based upon spectral data). To date, the system is still under revision by the Metabolomics Society. Going through the identification of all metabolites of interest is not realistic as the identification challenge is immense and confident unambiguous assignments of observed metabolic features to a single compound are not always possible (Warwick B. Dunn et al., 2013). Definitive (level 1) identification would require the comparison of two or more orthogonal properties (eg. rt, accurately measured m/z, fragmentation mass spectrum) of a chemical standard to the same properties observed for the metabolite of interest analysed

under identical analytical conditions. In this work, a level of confidence of 2 for metabolite identification (putatively annotated compounds) was reached. The application of accurate measurement of m/z was able to provide putative HMDB annotations (top 5 possibilities were considered) The Human Metabolite DataBase (HMDB, http://www.hmdb.ca) was used for mass-based metabolite annotation (Wishart et al., 2009). The HMDB is a web-based database that has been developed by The Metabolomics Innovation Centre (TMIC) to facilitate the


identification of metabolites using accurate mass data. It includes an annotated list of structural information for known metabolites and focuses on those found in human body fluid. These databases have links to other public databases such as the Kyoto Encyclopedia of Genes and Genomes (KEGG). Public database annotations were performed to the processed data sets with putative metabolites matching these queries. Annotations were carried out by matching the measured accurate masses ± 0.01 amu with theoretical ones.

Systems biology analysis A pair-wise analysis of de-regulated protein and metabolite was performed throughout the experiment. This pair-wise comparison of proteins or metabolite features is a representation of data where the individual values contained in the table were represented as colours. The range was set from -0.58 to 0.58 (Base 2 logarithm = 0.58), where < -0.58 was set to green, 0 to black and > 0.58 to red. The values in between are shown as colour gradients. The significantly different features are the ones that are lower than -0.58 and higher than 0.58. Identified proteins and annotated metabolites were analyzed using Ingenuity Pathways Analysis (IPA) (Ingenuity Systems®, Redwood City, CA, USA). Identified proteins were mapped onto Ingenuity’s Knowledge Database to generate networks on the base of their algorithmical connectivity. Canonical pathway analysis identified the

most significant ones from the IPA library, on the basis of the number of molecules from the data set that map the pathway. Functional analysis of networks revealed the biological functions most significant to the molecules in the network (p < 0.05, right-tailed Fisher’s exact test).

Trypan Blue assay For Trypan Blue assay, 5 x 104 cells were seeded in 3 mL complete culture medium


(transparent 6-well plate with flat bottom, Falcon, San Angel, Mexico) in 3 mL of complete culture medium for the 72 h of exposure times. Twenty-four h after seeding, cells were treated with 300 µM of 5 or 30 nm AuNPs. At the end of the exposure time, cells were washed with Phosphate Buffer (PBS, Invitrogen, Italy), detached with 0.5 mL of 0.05% Trypsin-0.02% sodium ethylenediaminetetracetic acid (Trypsin-EDTA, Invitrogen, Italy), and harvested with 1 mL of complete culture medium. A 30 µL of each sample was stained with Trypan Blue (Sigma-Aldrich, Italy) and counted with the TC10 automated cell counter (BIORAD, Italy) according to the supplier’s protocol. Negative control (untreated cells) was also included in the test. Data were reported as the mean of three independent experiments (3 replicates each) ± standard error mean and expressed as total cell number in respect to the negative control.

Immunocytochemistry analysis 5 x104 Caco-2 cells were seeded on a BD FalconTM Chamber polystyrene Vessel (Bedford, USA). Cells were exposed to AuNPs 5 and 30 nm at 300 µM for 72 h. At the end of the incubation time, cells were washed twice in PBS, fixed with 4% (v/v) paraformaldehyde and permeabilised with 0.05% (v/v) Triton-X (Sigma-Aldrich) in PBS (v/v) for 20 min at room temperature (RT). Staining was done with antibody against Annexin V (FL-319) (Santa Cruz

sc-8300 Cat. AB21679, Italy) diluted 1:250 in 3% (w/v) BSA in Tris-buffered saline (TBS). After removal of the primary antibody, cells were washed three times with 0.05% (v/v) Tween-20 in PBS and incubated secondary anti-rabbit IgG antibody labelled with Alexa Fluor® 546 (Lifetechnologies Cat. A11010). Alexa Fluor® 488 phalloidin (Lifetechnologies, Cat A12379, Italy) diluted 1:1000 was used for F-actin staining. VECTASHIELD® HardFSet™ Mounting Medium with DAPI (Vector Laboratories, Inc., U.S.A., Cat. H1500) was used to counterstain nuclei. Images were acquired with an Axiovert 200M inverted


microscope equipped with ApoTome slide module and Axiovision 4.8.2 software (Carl Zeiss; Jena, Germany), using 40 x/1.0 objective lens. For fluorescence microscopy quantitative analysis, images of the nuclei and cytoskeleton staining of Caco-2 cells exposed to AuNPs for 72h were acquired using INCell Analyzer (GE Healthcare Life Sciences, Belgium). Cytochalasin B (10 µg/mL) was used as a positive control for cytoskeleton disruption and apoptosis induction. Data analysis was then performed on the INCell Developer software (GE Healthcare, Life Sciences, Belgium) using in-house developed protocols. Cells were segmented using nuclei staining. The standard deviation of the pixel intensities within the nuclei was selected as a marker of apoptosis, and the standard deviation of the pixel intensities of phalloidin staining was used as a measure of cytoskeleton disruption. All data are expressed as the mean ± standard deviation of 20 fields acquired from three independent experiments performed in triplicate. Statistical analysis was performed using one way Anova, samples were considered significantly different from control when the p value was lower than 0.05 (95% confidence).

Apoptosis antibody array The human apoptosis antibody array-membrane (Abcam® - Italy, ab134001) was used for the simultaneous detection of human apoptosis markers according to the manufactures'

instruction (600 µg of cell lysate). Local background subtraction has been applied in the data analysis.

Figures and Tables (Supplementary Material)

Figure 1S. Selection of differentially expressed proteins for MS identification. Sypro Ruby


stained gel of cytoplasmic proteins extracted from untreated Caco-2 cells. The 61 selected protein spots are highlighted in circles and are marked with code numbers as specified in Table 1. Figure 2S. Cell viability using Trypan Blue. Representative images of Caco-2 cells incubated with or without 300 µM AuNPs of 5 or 30 nm for 72 h. The total cell number per well in each experimental condition was evaluated at the end of the exposure time using the Trypan Blue assay, as an indicator of cell proliferation inhibition. Figure 3S. Immunofluorescence staining of Caco-2 cells. Representative images of cells incubated with or without 300 µM AuNPs (5 and 30 nm) for 72 h and stained with A) DAPI to counterstain nuclei (DAPI, blue), B) Phalloidin (Alexa 488, green), C) Annexin V (DsRed, red). D) Merge images (A + B + C). E) Merge images (B + C). Scale bar: 20 µm. Figure 4S. Fluorescence microscopy quantitative analysis. Images of the nuclei (DAPI) and cytoskeleton staining (Phalloidin) of Caco-2 cells exposed to AuNPs for 72h were acquired using INCell Analyzer. Results of the data analysis for the nuclei area, apoptosis indicator and cytoskeleton disruption for control, 5 mn AuNPs, 30 nm AuNPs treatment and positive control are shown. Red arrows indicate fragmented nuclei. Scale bar: 50 µm. Figure 5S. Apoptosis antibody array. Cell lysates (600 µg) were incubated overnight with Abcam's human apoptosis antibody array-membrane. The antibody array membranes were then washed and a cocktail of biotinylated antibody mix was used to detect apoptosis-related

proteins. After incubation with HRP-conjugated streptavidin, the signal was visualized after 1 minute exposure by chemiluminescence. Comparison of signal intensity was used to determine relative differences in expression levels of the proteins. Positive control spots were used for normalization. Figure 6S. TEM images of as-synthesized AuNPs. Representative transmission electron microscopy (TEM) images of AuNPs as-synthesized. (A) 5 nm AuNPs; (B) 30 nm AuNPs.


Table 1S. Size distribution analyisis of the AuNPs. dCLS (nm) represents the position of the maximum peak of the AuNPs size based on weight distribution, calculated by CLS, σCLS (nm) is the size distribution by CLS calculated at Half Height Width (HHW) of the main peak. DICLS (nm) is the CLS Polydispersity Index. dDLS (nm) is the average size calculated by DLS, σDLS is defined as σDLS = dDLS (PDI)1/2. PDIDLS is the dimensionless polydispersity index. DTEM and σTEM are the diameter and the standard deviation of NPs calculated by TEM. Zpotential is a measure of electrostatic charges between particles which is correlated to the stability of AuNPs in solution. σZ is the standard deviation.