Authors: Mathilde Biola-Clier, Jean-Charles Gaillard, Thierry Rabilloud, Jean Armengaud, Marie Carriere
Categories: Article, TiO, inhalation, lung, nanoparticle, omics, phosphoproteomics, proteomics
Source: Nanomaterials
Doi: 10.3390/nano10020185
Authors: Mathilde Biola-Clier, Jean-Charles Gaillard, Thierry Rabilloud, Jean Armengaud, Marie Carriere
TiO2 nanoparticles (NPs) are one of the most produced NPs worldwide and are used in many consumer products. Their impact on human health, especially through inhalation, has been studied for more than two decades. TiO2 is known for its strong affinity towards phosphates, and consequently interaction with cellular phosphates may be one of the mechanisms driving its toxicity. In the present study, we used a phosphoproteomics approach to document the interaction of TiO2-NP with phosphoproteins from A549 human pulmonary alveolar epithelial cells. Cells were exposed to 21 nm anatase/rutile TiO2-NPs, then their phosphopeptides were extracted and analyzed using shotgun proteomics. By comparing the phosphoprotein content, phosphorylation status and phosphorylation sites of exposed cells with that of control cells, our results show that by affecting the phosphoproteome, TiO2-NPs affect cellular processes such as apoptosis, linked with cell cycle and the DNA damage response, TP53 being central to these pathways. Other pathways including inflammation and molecular transport are also affected. These molecular mechanisms of TiO2-NP toxicity have been reported previously, our study shows for the first time that they may derive from phosphoproteome modulation, which could be one of their upstream regulators.
Titanium dioxide (TiO2) is widely used in daily products as a white pigment, for example, as a food additive and in paints [1]. It is also used in cosmetics as a UV filter [1]. The annual production of nanosized TiO2 ranges between 3800 and 7800 tons in the US. With such a large production, TiO2-NPs could be present in 9% of nanomaterial-containing products [2].
The impact of TiO2-NPs on the lung has been the subject of intense research since inhalation is considered as being the first and most problematic route of occupational exposure. In vivo, exposure to TiO2-NP either via intratracheal instillation or inhalation causes pulmonary inflammation, fibrosis and emphysema-like response [1,3,4,5]. TiO2 is classified as possibly carcinogenic (2B group) via inhalation by the International Agency for Research on Cancer (IARC) [6]. In vitro, TiO2-NPs are endocytosed in a plethora of cell models [7]. When internalized in cells, TiO2 accumulates in endosomes and is transferred to lysosomes. Although there is no evidence of its accumulation in mitochondria, TiO2 induces oxidative stress in exposed cells by increasing the intracellular level of reactive oxygen species (ROS) and impairing the antioxidant cellular response [3,8]. This effect has been shown to be associated with the inhibition of the nuclear factor (erythroid-derived 2)-like 2 (NRF2) in BEAS-2B lung cells [9] and in mice kidneys [10]. TiO2 also causes oxidative damage to DNA without accumulating inside cell nucleus, except if the cell has undergone mitosis, one of the stages of which is the rupture of the nuclear envelope [11]. This genotoxicity is concomitant to impairment of the mitotic spindle assembly and function [12], decreased DNA repair activities [13,14] and cell cycle progression [15,16,17,18,19,20], suggesting that TiO2-NPs exert indirect primary genotoxicity [11]. Finally, TiO2-NPs have been shown to affect the autophagic process [21,22], and this property has been proposed as a possible therapeutic approach to treat cancer [23]. The main driver of these cellular effects is currently unknown and whether multiple factors are involved remains to be established.
Phosphate is essential for cell survival. Protein phosphorylation/dephosphorylation reactions, carried out by kinases and phosphatases, regulate the activity of almost all cellular processes [24,25] TiO2 shows high affinity for phosphate groups, due to its ion exchange abilities and ligand exchange behavior towards Lewis bases (for review, see [26]) and for this reason it has long been used to purify phosphoproteins [27,28]. Since TiO2-NPs heavily accumulate in cytoplasmic vesicles, they may adsorb phosphoproteins on their surface, sequester them in these vesicles, and consequently interfere with their proper function. Moreover, kinases and phosphatases are phosphoproteins; their sequestration on the surface of TiO2-NPs may hinder their role in maintaining the pool of cellular phosphoproteins. Therefore, we hypothesized that the impact of TiO2 on protein phosphorylation status could be one upstream mechanism driving its toxicity. This hypothesis is supported by some results showing impact of TiO2-NPs on cellular phosphoproteins. For instance, exposure of BEAS-2B cells to TiO2-NPs increases the phosphorylation level of several central kinases, including P38, c-Jun NH2- terminal protein kinases (JNKs), mitogen-activated protein kinase/extracellular signal-regulated kinase (ERK1/2), mitogen- and stress-activated protein (MSK1), glycogen synthase kinase (GSK), AMP-activated protein kinase (AMPK), signal transducer and activator of transcription (STAT) and the tyrosine protein-kinase FYN [29].
Consequently, the aim of this study was to characterize the impact of TiO2-NPs on protein phosphorylation profile in lung cells. Rather than using a case-by-case method, we used a large-scale approach, i.e., phosphoproteomics. This approach provides an exhaustive view of cellular phosphoproteome alteration in cells exposed to TiO2-NPs as compared to unexposed cells. We used a label-free phosphoproteomics approach consisting in enriching cellular phosphopeptides, then identifying and quantifying them by shotgun proteomics. This approach was applied to A549 human lung epithelial alveolar cells, exposed for 24 h to 100 µg/mL of 21 nm anatase/rutile TiO2-NP, i.e., 17 µg/cm^2^ equivalent to 0.11 ng/cell. This concentration corresponds to the alveolar deposition of TiO2-NP after inhalation exposure to 1 mg/m^3^ for a whole working lifetime [30]. Therefore, it can be considered as a worst-case exposure scenario. This exposure condition leads to moderate cell mortality, as inferred from our previous studies [14,31].
Unless otherwise indicated, chemicals were purchased from Sigma-Aldrich (Saint-Quentin Fallavier, France) and were >99% pure. NM105 TiO2-NPs were supplied by the Joint Research Center of the European Commission (JRC, Ispra, Italy). These NPs exhibit a spherical, ellipsoidal, and cuboidal structure, are 86% anatase and 14% rutile. Their diameter is 21 nm and their specific surface area 46 m^2^/g [32]. NPs were dispersed in ultrapure sterile water, by sonicating them in water for 30 min at 4 °C, using a high energy sonicator in pulsed mode (1 s on/1 s off), using an Autotune 750 W sonicator from Fisher Bioblock Scientific (Rungis, France) operated at 30% of amplitude, as previously described [14,31]. According to the calorimetric procedure, this corresponds to 19.82 W [33,34]. Suspensions were diluted in serum-free cell culture medium immediately before cell exposure. Their hydrodynamic diameter was 70 ± 20 nm with a 0.18 ± 0.04 polydispersity index (PDI) in water; it shifted to 720 ± 20 nm (PDI: 0.50 ± 0.02) after dilution in the exposure medium (mean ± standard deviation, n = 3, not shown).
This study was performed on A549 human epithelial alveolar cells (A549, ATCC CCL-185), grown in DMEM as previously [14]. The use of this cell line in toxicity studies has been criticized because it exhibits a constitutively active NRF2 [35]. However, we chose to use it in the present study because to our knowledge it is the only available cell line of human alveolar origin and NPs have been shown to deposit in this region of the lung due to their small size [36]. They were exposed, 3 days after seeding, to 100 µg/mL of TiO2-NPs diluted in serum-free cell culture medium, with three biological replicates per condition (n = 3), then rinsed three times with PBS and harvested using trypsin. This concentration corresponds to 17 µg TiO2/cm^2^, i.e., 0.11 ng TiO2/cell. At this concentration, these NPs cause approximately 25% of cell viability loss (as shown in [31], where these NPs are referred to as “TiO2-Degussa” and in [13]), but significantly increases intracellular reactive oxygen species level, causes DNA damage and significantly increases the DNA repair activities in A549 cells (as shown in [14], in this article these NPs are referred to as “A25”, and in [13]). Cell exposure was performed in the dark in order to avoid any photocatalytic effect of TiO2-NPs.
The phosphoproteomics experiment was performed with 1.2 × 10^7^ cells per replicate, which were sampled using trypsin and flash frozen in liquid nitrogen. Samples were stored at −80 °C until analysis. The samples were melted on ice in a lysis buffer consisting in 50 mM Tris/HCl, pH 8.2, 8 M urea, 75 mM NaCl and 2X phosphatase inhibitor cocktail (Thermo Scientific, Illkirch, France, 88667). They were then sonicated three times using a Hielsher (Teltow, Germany) UP50H probe sonicator, operated in pulse mode (20 s at 0.4 s on/0.4 s off) at 40% amplitude, with a pause of 1 min between each cycle of sonication. After centrifugating these lysates at 16,000 × rcf for 10 min, 4 °C, the supernatants were stored at −80 °C. After thawing, protein concentration in the samples was measured using Bradford reagent (Interchim, Montluçon, France, CooAssay Standard Protein Assay kit), using the protocol of the supplier. The reduced cysteine residues were alkylated by reaction with iodoacetamide (5 mM), for 15 min at room temperature. After dilution with 100 mM Tris/HCl, pH 8.5, to reach an urea concentration of 1 M, the samples were digested overnight at 37 °C using 1 mg/mL trypsin prepared in a 0.01% aqueous solution of trifluoroacetic acid, so that the final concentration of protein was 2% per sample. After centrifugating for 1 min at 1000 rcf, at room temperature, the supernatants were stored at −80 °C. SDS-PAGE followed by Coomassie blue staining was used to evaluate the extent of proteolysis.
Digested samples were desalted using Harvard Apparatus (Holliston, MA, USA) Macro Spin C18 Columns, using the protocol of the supplier, consisting in (i) activation of the resin in 500 µL of 80% acetonitrile, 20% water, repeated three times, (ii) equilibration of the columns by rinsing with 500 µL 0.5% trifluoroacetic acid, three times. 250 µg of digested peptides was loaded onto columns, which were then centrifuged at room temperature for 30 s at 2000 rcf. The filtrates were then collected and loaded again onto the columns. At the end of this procedure, the resins were rinsed three times with 0.5% trifluoroacetic acid and peptides were eluted in 100 µL of 1 acetonitrile/water, 0.1% TFA. These desalted peptides were freeze-dried and stored at −80 °C.
As described in Figure 1, the samples were then subjected to two consecutive chromatographies. The first one was a strong cation exchange chromatography (SCX), which was processed as previously described [37], using a polysulfoethyl ATM column (PolyLC, Columbia, MD, USA, 3.2-mm inner diameter, 200 mm length, 5-µm particle size, 200 Å pore size) operated at a flow rate of 400 µL/min on a 1100 Series reverse-phase high performance liquid chromatography (HPLC) system equipped with a G1315B diode array detector, a G1322A degasser and a G1311A quaternary pump, as well as a G1329A autosampler, and sample cooler unit from Agilent (Les Ulis, France) [38]. Samples (1 mg of peptide) were diluted in 65 µL of buffer A (7 mM KH2PO4, pH 2.65 and 30% acetonitrile (vol./vol.)) and 60 µL was injected in the column. The stepwise elution was performed with three buffer A, buffer B (7 mM KH2PO4, 350 mM KCl, pH 2.65, 30% acetonitrile (vol./vol.)), buffer C (50 mM K2HPO4, 500 mM NaCl, pH 7.5) and ultrapure water. Buffer A was applied to the column from time 0 to min 34. Then we applied 75% buffer A and 25% buffer B for one min, then buffer B from min 36 to 41, then ultrapure water from min 42 to 48. Buffer C was then applied from min 49 to 60, followed by ultrapure water from min 61 to 67. Finally, Buffer A was applied from min 68 to 120. Phosphopeptide samples were collected as two separate fractions of 4 min-elution, between min 8 and min 16, as described previously [37]. These two separate fractions were freeze-dried in order to reach a final volume of 150 µL, to which was added 350 µL of 0.1% TFA. The samples were desalted as previously. The second chromatography was used to enrich phosphopeptides, and consisted in immobilized metal affinity chromatography (IMAC). The resin was the PhosSelect Iron Affinity Gel resin (Sigma-Aldrich, Saint-Quentin Fallavier, France, P9740), to which samples were bound as previously described [37]. Peptides were then desalted as described above, except that samples were rinsed with 200 µL instead of 500 µL. To do so, the resin was loaded on activated and equilibrated Harvard Apparatus (Holliston, MA, USA) Micro Spin C18 columns. Therefore, we obtained a two-layered resin, which was rinsed twice with 100 µL of 40% acetonitrile, 60% 25 mM formic acid (IMAC binding buffer) and once with 200 µL of C18 wash solution. Phosphopeptides were eluted from the PhosSelect resin and retained on the underlying C18 resin by adding three times 100 µL of 500 mM K2HPO4, pH 7 (IMAC elution buffer). The resin was rinsed four times with 200 µL of C18 wash solution and phosphopeptides were eluted twice using 50 µL of C18 elution solution. These two eluted fractions were pooled. Samples were then freeze-dried and stored at −80 °C.
Phosphopeptides (10 µL) were loaded on the LTQ Orbitrap XL hybrid nano-liquid chromatography coupled to tandem mass spectrometry (nanoLC-MS/MS) system (ThermoFisher Scientific, Les Ulis, France) coupled to Ultimate 3000 LC system (Dionex-ThermoFisher Scientific, Les Ulis, France), after dissolution in 30 µL of 0.1% TFA. The analysis conditions were as described by Dedieu et al. [39]. Full-scan mass spectra were measured from m/z 300 to 1800. The mass spectrometer was operated in data-dependent mode using a TOP3 strategy consisting in a scan cycle initiated with a full scan of high mass accuracy in the Orbitrap (30,000 resolution; internal calibration), followed in parallel by MS/MS normal mode scans in the linear ion trap on the three most abundant precursor ions. The parameters were set as minimum signal 15,000; possible charge 2+ and 3+; dynamic exclusion of previously-selected ions with 60 sec exclusion duration. The previously described Multi Stage Activation mode (pseudo-MS3) [40] was activated with a neutral lost mass list including 32.6590 (phosphate, triple positive charges) and 48.9890 (phosphate, double positive charges). The lock mass option on the LTQ Orbitrap XL mass spectrometer was enabled in MS mode and the polydimethylcyclosiloxane ions generated in the electrospray process from ambient air (protonated [(CH3)2SiO)]6 with m/z at 445.12002) were used for internal recalibration in real time.
Peak lists were generated with the MASCOT DAEMON software (Matrix Science, London, UK, version 2.3.2) using the extract_msn.exe data import filter from the Xcalibur FT package (version 2.0.7) (ThermoFisher Scientific, Les Ulis, France). The filtering options were, as previously [41]: minimum 400, maximum 5000, grouping 0, intermediate 0, minimum 10, extract MSn: 2 and 1000. Peptides were assigned from MS/MS spectra according to the SwissProt database with the MASCOT 2.3.02 software (Matrix Science, London, UK), by searching the database with the following Mammalia taxonomy (65,476 sequences), SwissProt_2012_02, maximum number of miss-cleavages: 2; mass 5 ppm (parent ion) and 0.5 Da (MS/MS); carbamidomethylated Cys: fixed modification; and oxidized Met and phosphorylated Ser, Thr and Tyr variable modification. MASCOT results were analyzed using IRMa 1.30.4 [32] software, which filters out peptides with a p-value of less than 0.01 and a rank set to 1. Fold-changes represent the ratio of values obtained in samples exposed to TiO2 and samples from control cells, based on spectral counts after standard normalisation. PatternLab software was used for normalization and statistical significance assessment [42]. The cut-off values to consider results as statistically significant was p < 0.05 and fold-change > 1.5. Gene ontology and pathway analyses were performed using Database for Annotation, Visualization and Integrated Discovery (David) [43,44] and Ingenuity^®^ Pathway Analysis (IPA^®^—23814503, QIAGEN, Courtaboeuf, France). Enriched pathways were considered statistically significant when p < 0.05.
Cells were exposed to 100 µg/mL of TiO2-NPs, then rinsed three times with PBS containing 2 mM of EDTA. Then they were fixed for 30 min in 70% ice-cold ethanol and diluted in PBS-EDTA (2 mM). This fixative solution was removed and cells were suspended in 25 µg/mL propidium iodide prepared in PBS-EDTA (2 mM) to which was added 25 µg/mL RNase A. Samples were analyzed by flow cytometry using a FACS Calibur (Becton Dickinson, Rungis, France). This experiment was repeated three times independently, with four replicates per experiment.
Cells exposed to 100 µg/mL TiO2-NPs for 24 h were washed three times with phosphate saline buffer, fixed with 2.5% glutaraldehyde and post-fixed with OsO4. They were then dehydrated by immersion in solutions of ethanol with increasing concentration and embedded in Epon. Ultrathin sections were prepared by ultramicrotomy and counterstained with uranyl acetate and lead citrate. These samples were observed using a CM12 Philips electron microscope, operating at 80 kV.
Cell exposure conditions were chosen based on results that we previously obtained, showing mild cytotoxicity of TiO2-NPs to A549 cells, i.e., less than 25% of cell mortality in exposed cells according to the MTT assay. Moreover, we previously showed that this exposure condition leads to a significant elevation of intracellular reactive oxygen species level, suggesting oxidative stress. We also showed that it causes significant DNA damage, as assessed via the comet assay and quantification of 8-oxo-dGuo using high performance liquid chromatography coupled to tandem mass spectrometry (HPLC-MS/MS), as well as elevation of DNA repair capacities in exposed cells [13,14]. The same conclusions were obtained on BEAS-2B cells, which is a normal bronchial cells [13].
Phosphoproteomes of control cells and cells exposed to TiO2-NPs were analyzed after enrichment by high-resolution tandem mass spectrometry [37]. As shown in Figure 1, phosphopeptides in both non-exposed and exposed cells were enriched for each of the three biological replicates, resulting in two separate fractions (#1 and #2) per sample. These two samples were analyzed separately over a 90-min gradient by nanoLC-MS/MS with a high-resolution mass spectrometer, and the results were merged (Supplementary Table S1).
Among the recorded spectra for the three biological replicates 10,472 and 11,754 were assigned to human peptides in control cells and cells exposed to TiO2-NPs, respectively, corresponding respectively to 1310 and 1283 unique peptides (Supplementary Table S1). Among them, 88–90% were they mainly carried one phosphorylation (66–70%), but sometimes two (18–24%) and more rarely three (0.2%). This corresponds to 1606 unique phosphorylation sites (P-sites), which are listed in Supplementary Table S2. These peptides belong to 644 and 649 proteins in control cells and cells exposed to TiO2-NPs, respectively, 96% and 93% of them being phosphorylated, i.e., 606 phosphoproteins in control cells and 617 phosphoproteins in cells exposed to TiO2-NPs that are reported in Table S3. Among these proteins, 510 were detected in both control cells and cells exposed to TiO2-NPs (Supplementary Table S3, Figure 1).
We compared the protein phosphorylation level in control cells and in cells exposed to TiO2-NPs using three criteria. First, we compared the phosphorylation counts, i.e., the numbers of phosphorylated residues detected on all the peptide sequences taking into account the Peptide-to-Spectrum Matches. The second criterion was the phosphopeptide count, i.e., the number of spectra of phosphopeptides detected in the samples. The third criterion was the phosphorylated sites count, i.e., the number of unique phosphorylated sites on the detected peptides (see Figure S1 for a schematic representation). Using these three criteria, when analyzing the list of phosphopeptides detected in the samples, normalized with respect to individual protein molecular weights, the overall phosphorylation level did not statistically differ in CTL cells, compared to cells exposed to TiO2-NPs (Figure S2). When using these three criteria to analyze the phosphorylation level of each individual phosphoprotein, we identified significant differences between control cells and cells exposed to TiO2-NPs. The fact that these two complementary analyses led to opposite phosphorylation trends suggests that many phosphorylatable sites (P-sites) showed different phosphorylation level rather than a strong modification of phosphorylation level of a few P-sites (here, phosphorylatable sites are considered as those that have been identified phosphorylated at least once in a biological sample in the experiment). As reported in Figure 2, out of the 510 phosphoproteins identified in both control and TiO2-exposed cells, 89 proteins (17.5%) showed different phosphorylation levels in cells exposed to TiO2-NPs as compared to control cells according to at least one of the three criteria (Category 1). Twenty-six proteins showed different phosphorylation levels according to all three criteria (Category 2). Most of these phosphoproteins showed increased phosphorylation levels in cells exposed to TiO2-NPs, compared to control cells (Figure 2).
When using the same approach on individual P-sites, out of the 1606 detected P-sites, 139 showed different phosphorylation levels in control and TiO2-exposed cells. These 139 P-sites belonged to 111 phosphoproteins (Category 3). Contrary to what was found when analyzing protein phosphorylation level, 70% of P-sites were less phosphorylated in TiO2-NP-exposed cells than in control cells. Therefore, the overall protein phosphorylation level is not affected in cells exposed to TiO2-NPs. Considering the strong affinity of TiO2 towards phosphate groups, and the intense accumulation of these TiO2-NPs in A549 cells (see [31], where these TiO2-NPs are referred to as TiO2-Degussa), a much more intense dysregulation of the overall protein phosphorylation levels was expected. Our results suggest that the impact of TiO2-NPs on the cellular phosphoproteome is much more finely tuned. This could be explained by the cellular distribution of TiO2-NPs in A549 cells, which shows preferential location in endosomes and/or lysosomes [31]. Sequestration in such compartments would limit the contact of TiO2 with phosphoproteins.
The list of phosphocounts, phosphopeptides and phosphosites showing modulated phosphorylation level (p < 0.05) was analyzed via protein ontology, using David and Ingenuity^®^ Pathway Analysis (IPA, Qiagen, Courtaboeuf, France), followed by data mining using the Uniprot protein annotation. The whole human proteome was used as reference group. Since both David and IPA analyses, based on the lists of individual P-sites, phosphocounts or phosphopeptides, highlighted the same cellular functions and processes, we present here the results obtained using David and the analysis performed on the list of phosphopeptides, showing a false discovery rate (FDR) lower than 20% (Table 1). All GO terms identified on the list of phosphopeptides are reported in Table S4.
The list of GO terms obtained when analyzing phosphocounts (cluster analysis) and P-sites are reported in Supplementary Tables S5 and S6, respectively. The GO terms obtained via IPA, in the pathways “Diseases and Biofunctions”, for proteins listed in category 1, 2, and 3 are reported in Supplementary Tables S7–S9, respectively. For most of these proteins, the link between the phosphorylation status and their function is not established. Therefore, here, we report the overall function of these phosphoproteins, and if available, the role played by phosphorylation in their function.
Among the GO terms identified by David, “apoptosis” (FDR 6.95%) and “negative regulation of extrinsic apoptotic signaling pathway” (FDR 18%) attracted our attention, since apoptosis has already been described as being one of the cellular processes induced by TiO2-NPs [1]. The proteins identified here and involved in these pathways are EPHA2, BAG6, ACIN1, HTT, BAG3, MAP1S, SQSTM1, TPX2, CTTN, SON, KRT18, ADAR, and LYRIC. EPHA2 participates in UV-induced apoptosis [45] and ACIN1 induces chromatin condensation [46]. BAG3 exhibits anti-apoptotic activity when bound to BCL-2 [47]. TPX2 participates in the assembly of microtubules during apoptosis [48]. Some of these proteins also play roles in the autophagic process, which is also known as being dysregulated in cells exposed to TiO2-NPs [22,49,50]. For instance, MAP1S activates autophagy and consequently reduces cell propensity to undergo apoptosis [51]. HTT is implicated in autophagic vesicles formation [52] and it is phosphorylated in response to DNA damage; its phosphorylation is linked with its toxicity [53]. SQSTM1, when linked to p62, constitutes a selective autophagy receptor that directs ubiquitinated substrates to degradation [54], while SPP1 regulates ER stress-induced autophagy [55] and RB1 is also involved in autophagy [51]. In line with this observation, we previously showed that repeated exposure of A549 cells for 2 months to the same TiO2-NPs as in the present study leads to significant accumulation of TiO2-NPs in autophagic vesicles and increased LC3II/LC3I ratio, suggesting alteration of the autophagic process [49]. This alteration would result from modulation of the phosphorylation of these proteins. Moreover, some of these proteins play roles in the DNA damage response, which is activated in TiO2-NP-exposed cells, particularly via impairment of the DNA repair as previously demonstrated in A549 cells exposed to the same TiO2-NPs and in the same conditions as in the present study [13,14] and via modulation of cell cycle progression [15,16,17,18,19,20]. This is the case of BAG6, which is involved in DNA-damage induced apoptosis by binding EP300, itself involved in the regulation of TP53 transcriptional activity [56]. Moreover, phosphorylated ACIN1 modulates the expression of cyclin A1 that is involved in the control of cell cycle progression, particularly in the transitions between G1 and S phase and between G2 and M phase [57]. SON is involved in splicing of many DNA-repair transcripts, and is required for the progression of the cell cycle [58]. In addition, ITGAV is involved in TGF-β1 activation [59], while RB1 regulates the entry into cell division. It is either active or inactive depending on its phosphorylation status. Depending on its phosphorylation level, it is also involved in TGF-β1-induced apoptosis [60]. Moreover, NDRG1 shows different phosphorylation level in cells with active cell cycle or with blocked cell cycle [61,62]. Its phosphorylation level controls its nuclear localization and its role in cell division, DNA repair and proteasomal degradation [62]. Analysis of the cell cycle in A549 cells exposed to 100 µg/mL of TiO2-NPs for 24 h shows a mild but statistically significant reduction of the proportion of cells in the G1 phase, as compared to unexposed cells (Supplementary Figure S3). Moreover, after repeated exposure of A549 cells to the same TiO2-NPs as in the present study, we previously observed decreased cell proliferation and perturbation of the cell cycle [49,63]. Moreover, a significant increase of the proportion of cells in the sub-G1 phase is observed, implying that some cells undergo apoptosis (Supplementary Figure S3). This suggests that this effect on the cell cycle, visible at the scale of the phosphoproteome, translates into a phenotypic effect, which is revealed after acute and prolonged exposure.
Our analysis highlights the central role of TP53 in these processes, as illustrated by a network identified using IPA (Figure 3). This network focuses on cellular assembly and organization, DNA replication, recombination and repair, and cancer pathways. The phosphorylation of TP53 itself is not shown to be affected, but p53 acts as an upstream transcriptional regulator of several proteins having modulated phosphorylation levels (SKIV2L, TPX2, BAG6, SON, KIF4A, POP1, RPLP1, DERL1, MARK2, ZC3HC1, PHC3). We previously showed that repeated exposure of A549 cells to the same TiO2-NPs as in the present study, up to two months, leads to increased phosphorylation and acetylation of TP53 [49]. This suggests that the TP53 pathway is activated in A549 cells exposed to these TiO2-NPs. Induction of DNA damage and/or apoptosis is known to be related to activation of the TP53 pathway [64,65,66,67,68,69], we show here that these cellular processes are at least partly impacted via dysregulated phosphorylation.
Impaired cell cycle progression induced by TiO2-NPs is generally reported as cell accumulation in S or G2-M phases [15,16,17,18,19,20]. It is concomitant with impairment of mitotic progression, formation of multipolar mitotic spindles, abnormal chromosome segregation during anaphase and telophase, with deregulation of the function of PLK1 kinase [12]. The present analysis suggests that dysregulated phosphorylation may be a mechanism through which all these processes are affected.
Several pathways identified by David-based analysis are related to disturbance of the cytoskeleton, both actin cytoskeleton and microtubules (“Cytoskeleton” FDR 0.05%, “actin cytoskeleton” FDR 0.52%, “actin binding” FDR 2.89%, “microtubule cytoskeleton organization” FDR 6.27%, etc.). Some of these proteins play roles in cell migration, motility, and adhesion. For instance, TLN2 is a component of focal adhesion plates and is involved in cell adhesion in synapses, ZYX is a stabilizer of focal adhesions in muscle cells and synapses [70], and CTTN is involved in cell migration and the formation of metastases [71]. Other proteins rather play roles in cellular structure, such as MAP1B, which is involved in microtubule polymerization and stabilization, in the formation of autophagosomes and in membrane blebbing [72]. Likewise, MPRIP stabilizes actin fibers and plays a role in development of stress fibers [73] and KIF4A is necessary for successful cytokinesis [74]. Moreover, some of these proteins are involved in intracellular trafficking, such as NDRG1, which is involved in trafficking of vesicles and lipids and CTTN, which is involved in protein trafficking and in endocytosis [75]. This impact on the cytoskeleton is illustrated by a second network identified using IPA, which involves 35 proteins and illustrates the impact of TiO2-NPs on cellular development, movement and morphology (Figure 4).
Transmission electron microscopic observation of these cells exposed to TiO2-NPs showed strong accumulation of NPs inside cells, and localization in cytoplasmic vesicles (Figure 5). This accumulation did not lead to any change in the morphology of cells, and we did not detect the presence of stress fibers (Figure 5).
The impact of TiO2-NPs on cytoskeleton integrity highlighted in this phosphoproteome analysis is consistent with the literature [17,76,77,78,79,80,81,82,83,84]. Indeed, TiO2-NPs have been shown to impair the expression of genes involved in cytoskeleton maintenance [17,81]. It has also been shown to induce disorganization of microtubules and of the actin network [76,79,80,83]. Moreover, in an acellular study, TiO2-NPs were also shown to inhibit tubulin polymerization and change its conformation [77]. Again, we show here that dysregulated phosphorylation of some phosphoproteins may be a mechanism leading to this impact.
In addition, several pathways identified in this gene ontology search are related to RNAs and their processing (“poly(A) RNA binding” FDR 0.03, “alternative splicing” FDR 0.25, “RNA-binding” FDR 19.65), particularly to splicing of pre-mRNA and mRNA (SON, SNRNP200, SRRM1, ADAR, ACIN1). SON is a cofactor for transcript splicing, especially for transcripts related to cell cycle and DNA repair [85], while SNRNP200 is involved in spliceosome assembly and activation and it is a putative helicase [86]. SRRM1 is part of the mRNP granules, which regulate mRNA translation, localization, and recycling [87]. ADAR is involved in the editing of double-stranded RNA [88], which affects mRNA translation. ACIN1 is part of the exon junction complex, which is involved in mRNA processing [89]. One of these proteins is a component of the MRP ribonuclease complex and of ribonuclease P, which is involved in the generation of mature tRNA and (POP1). Finally, one of these proteins is a helicase involved in the biogenesis of 60 S ribosomal subunits (DDX51). This suggests a possible impact of TiO2-NPs on translation, which would be mediated by dysregulated phosphorylation of some phosphoproteins. RNA-binding proteins have been reported to bind to SiO2 nanoparticles, and consequently to affect translation [90]. The present results suggest that TiO2-NPs could also affect translation through modulation of the phosphorylation of proteins involved in RNA processing. This would explain the overall gene expression downregulation that we previously observed in A549 cells exposed to the same TiO2-NPs, and in the same exposure conditions [13].
Finally, several proteins are involved in molecular transport (ADAR, CTTN, HTT, NUP214 and SEC61B), in and out of the nucleus in particular. NUP214 is a nucleoporin involved in nuclear export [91] and SEC61B is responsible for the nuclear import of proteins, in particular the epidermal-growth factor receptor (EGFR) [92]. Moreover, SQSTM1 also shows dysregulated phosphorylation level. It has been reported that chromosomal translocation leads to the production of a NUP214-SQSTM1 fusion protein that forms nuclear bodies, which shuttle between the nuclear and cytoplasmic compartments, and may be a form of storage of nuclear transport proteins [93]. Dysregulated phosphorylation levels of NUP214 and SQSTM1 might be the result of dysregulated levels of the single proteins, or of the fusion protein. Modulation of the phosphorylation profile of these proteins might affect the nuclear localization of some key proteins, for instance, some proteins involved in DNA repair, explaining the accumulation of DNA caused by TiO2-NPs [11] or impaired DNA repair activities [13,14]. These results are consistent with recent studies, which report that TiO2-NPs dysregulate the expression of genes or proteins related to ion transport [81,84], trans-membrane transport [94,95], and molecular transport inside the cell [29,96].
We then analyzed the literature relative to the function of the 139 dysregulated P-sites, which would explain some of the impacted biological activities. The biological function of only 14 P-sites is known (Table 2). Some kinases responsible for the phosphorylation of these P-sites have been reported, while phosphatases are known only for two of them, namely CTTN^421^ and CARHSP^141^, which are dephosphorylated by PTP1B and PP2A, respectively [78,97]. Again, the two main functions that are impacted via these dysregulated P-sites are cell cycle regulation and cytoskeleton organization. AKAP^1151^, ENSA^67^, ERF^526^, FLNA^1084^, KIF4A^801^, MISP^397^ and RB^1826^ have been shown to regulate the cell cycle [98,99,100,101,102,103,104], while AKAP^1151^, CTTN^421^, FLNA^1084^, HSPB^115^ and VIM^5^6 have been reported to be involved in cytoskeleton organization [97,98,103,105,106,107,108,109].
Other functions are identified through this approach, i.e., the oxidative stress response (HSPB^115^ and CARHSP^141^) [105,108,110,111,112] and inflammatory processes (VIM^56^ and NDRG^1330^) [107,113]. Both oxidative stress and inflammation are hallmarks of TiO2-induced cellular stress [1]. We show here that dysregulated phosphorylation of some key proteins may at least partly drive these effects.
Analysis of the phosphoproteome of A549 cells exposed to TiO2-NPs reveals that these nanoparticles affect DNA damage response, by means of dysregulation of the cell cycle and DNA repair, the process of autophagy, the cytoskeleton dynamics and structure, the RNA dynamics, as well as intracellular transport. It highlights the central role of p53 in some of these processes. Although these effects of TiO2-NPs have already been reported using other techniques, here we suggest that there is a link between the dysregulation of phophoprotein phosphorylation status and these cellular events, using an exposure condition that does not induce any overt cytotoxicity. It should be highlighted that the A549 cell line is derived from an alveolar carcinoma. Therefore, it would be important to repeat this experiment on a non-cancerous cell line. This would demonstrate that the observed effects are not specific to cancer cells and that they are representative of the impact caused by inhalation exposure of healthy individuals to TiO2-NP.