*3.3. Pathways Analysis in Response to CdS QDs Treatment*

Several pathways were identified by Mapman ontology as being particularly involved in the response to CdS QDs. In wt and both mutants, proteins from cellular metabolism such as glyceraldehyde-3-phosphate dehydrogenase (Gapc1, Gapb, Gapa1 and Gapa2), pyrophosphate fructose6-phosph1-phosphotransferase subunit beta1 (Pep-beta1), 30S ribosomal protein S5, glycine dehydrogenase decarboxylating 2 (Glpd2) and bifunctional enolase 2/transcriptional activator (Eno2) were modulated (Figures 2 and 3 and Table S2). Arabidopsis has four different GAPDH isoforms, with seven phosphorylating types and one non-phosphorylating. These include cytosolic glycolytic GAPDHs (GAPC1 and GAPC2), chloroplastic photosynthetic GAPDHs (GAPA1, GAPA2, and GAPB), plastidic glycolytic GAPDHs (GAPCp1 and GAPCp2), and the NADP-dependent non-phosphorylating cytosolic GAPDH (NP-GAPDH) [50]. Substrate conversion by glycolytic GAPDHs catalyzes a simultaneous reduction of NAD+ to NADH [51]. Arabidopsis GAPA1/2 and GAPB use NADPH to generate NADP+, which buffers free radical generation from the electron transport chain by dissipating the H+ gradient in the thylakoid membrane [52,53]. Therefore, by contributing to the maintenance of the NAD(P)+/NAD(P)H ratio of the cell, plant GAPDHs influence cellular redox and general metabolism. In particular, Gapc1 plays a role in the glycolytic pathway, but at the same time, it can interact with H2O2 thus becoming part of the ROS signaling cascade [51]. The regulation of several proteins involved in primary metabolism suggests that CdS QDs exposure had moderately influenced carbon metabolism. These include two proteins involved in the glycolytic pathway: the upregulation in wt of Gapc1 and the downregulation in two mutants of Eno2 (Figures 2, 3 and 5 and Table S2). This observation further supports the concept that CdS QDs exerts oxidative stress by reacting with cellular proteins and enzymes and subsequently generated free radicals [54]. As hypothesized by Tiwari et al. [55], the disruption or malfunction of the electron transport system in mitochondria and chloroplast by QDs, once they are inside the cell, could lead to ROS production [55]. In wt and the two mutants many proteins are involved in mitochondrial electron transport including V-type proton ATPase subunit B1 and B2 (Vha-b1, Vha-b2), prohibitin-3 (Phb3), and gamma carbonic anhydrase 2 (Gammaca2), with, in the chloroplast Calvin cycle, phosphoribulokinase (At1g32060) and chaperonin 60 subunit beta 1 (Cpn60b1) (Figure 3 and Table S2). It has been demonstrated that physical interaction with ENMs disrupts the normal function of organelles in cells. For example, mitochondrial damage is thought to be one of the possible mechanisms of ENMs cytotoxicity by inducing oxidative stress through the destruction and redistribution of normal electron transport by respiratory complexes [56]. V-type proton ATPase subunit B1 was downregulated in wt and in *atnp*01, while V-type proton ATPase subunit B2 was downregulated in *atnp*02 (Figures 2 and 5). A study on Arabidopsis mitochondrial proteomics identified enolase among the enzymes associated with the outer mitochondrial membrane [57], and enolase is also proven to interact with the tonoplast through direct association with V-ATPase subunits, specifically the regulatory subunit Vha-b [58]. According to Ruotolo et al. [15], CdS QD treatment decreases respiratory efficiency and chlorophyll content in Arabidopsis [15]. Therefore, our findings support the involvement of the mitochondria respiratory process in CdS QDs responses in wt and both mutants. Interestingly, in the two mutants, respiration and photosynthesis were less sensitive to QDs than wt. Most of the studies carried out so far have dealt primarily with the overall plant stress response towards specific groups of ENMs, showing a differential abundance of proteins involved in oxidation-reduction, reactive oxygen species (ROS) detoxification, stress signaling, and hormonal pathways [59].

**Figure 5.** Model depicting the main A. thaliana proteome pathways responsive to CdS QDs treatment. For each protein the colored bars indicate up- (red) or downregulation (green). The first column indicates the up- and downregulated proteins in wt, while the second and third columns similarly show those in mutants *atnp*01 and *atnp*02.

Interestingly, five glutathione S-transferases (GSTs), representative of λ and φ classes of the GST family, have been identified in the proteome analysis (Figure 2 and Table S2). In wt, two GST family members, Gst F2 and F9, decreased under treatment, while GstF7 and Gst-darh3 increased; conversely, in *atnp*02, GstF3 was upregulated (Figures 2 and 5). The GST family in plants is notable for its structural and functional diversity, but the biochemical and physiological functions of each specific member remain to be clarified. As well as, or instead of, catalyzing conjugase reactions, some GSTs have antioxidative functions. The DHAR type of GST is one example. In addition, several subclasses of GST have peroxidase activity [60,61]. The detection of a large number of GSTs in the A. thaliana proteome is quite interesting; the functions of GSHs in the maintenance of the cell's redox balance, in xenobiotic detoxification and flavonoids subcellular transport is well known. GST expression is induced by a broad swathe of stressing conditions [62], including ENMs [63,64]. In Arabidopsis, ENMs induce an oxidative stress response by producing ROS, as has also been reported in crop plants [64,65]. Genes encoding for proteins of the NADPH oxidase and superoxide dismutase (SODs) families, and particularly peroxidases (PODs) and gluthatione S-transferase (GST) families, involved in antioxidant pathways that promote ROS detoxification, become significantly modulated under CdS QD and CuO and ZnO NP treatments [24,66,67].

Exposure to ENMs may result in the change of transcription of genes involved in hormone signaling pathways, e.g., of auxin repressor or auxin response genes, abscisic acid (ABA) biosynthetic genes, or ethylene signaling components [68]. The present study, shows that in A. thaliana wt CdS QDs caused upregulation of the protein involved in the biosynthesis of auxins (auxin-responsive protein IAA9); in *atnp*01 there was downregulation of proteins involved in the biosynthesis of ABA (molybdenum cofactor sulfurase), and in *atnp*02 there was upregulation of a protein involved in the biosynthesis of jasmonate (JA) (PYK10-binding protein 1) and downregulation of the protein involved in the biosynthesis of auxin (IAA9) (Figures 2 and 5 and Table S2). ABA plays a key role in lateral root formation inhibition in plants exposed to environmental stress [69], and it is also an antagonist of brassicosteroids-promoted growth. Genes induced by ABA are upregulated by CdS QDs [14], CuO [66], and ZnO [67] NPs. In plants, protein phosphatase 2C 58 (PP2C) has been connected with the negative regulation of protein kinase cascades that are activated because of stress. Members of the PP2C family, such as ABI1 and ABI2, are involved in ABA signal transduction. ABA is a plant hormone crucial in mediating the

plant responses to environmental stresses [70]. In this work, we found that ABI1 and ABI2 are upregulated in the wt and downregulated in both mutants when treated with CdS QDs (Figures 2 and 5). These proteins are downregulated in both mutants probably because the signal transduction carried out by the hormone ABA is impaired in both mutants by the nanoparticles. The major role of JA (jasmonic acid) is defense against pathogens. However, this hormone plays a role in plant growth control [71]; the transcription of some JA responsive genes increased upon exposure to certain ENMs. Several genes involved in SA (salicilic acid) pathways, a signaling molecule that is important in general plant stress response, were downregulated by early exposure to Ag and TiO2; however, they were upregulated by exposure to CdS QD, CuO, and ZnO NPs [14,66,67]. The modulation of these proteins by CdS QDs treatment highlights the importance of hormones and signaling in the response mechanisms to these nanoparticles. Figure 5 gives an overview of all the above-mentioned pathways responsive to treatment with CdS QDs.

#### *3.4. Network Analysis*

Starting from the molecular structure of the two mutants *atnp*01 and *atnp*02, and using a network analysis, we tried to find any among the identified proteins which showed variations that could be correlated with the mutations. Three candidate genes were potentially affected by the mutagenic events in *atnp*01 [14]. Two of these genes (At1g13880 and DLR1) are positioned on chromosome 1, the other (At3g46880) is on chromosome 3 (Figure 6) and encodes an unknown protein localized in the chloroplast, which does not interact with any other protein. DRL1 encodes a cytoplasm-localized calmodulin binding protein involved in leaf development and meristem structural organization. DLR1 is a homolog of the yeast TOT4/KTI12 protein which associates with elongator, a multisubunit complex binding to RNA polymerase II transcription elongation complex [72]. DRL1 interacts with two differentially abundant proteins found with 2D-PAGE: elongator complex protein 3 (Elo3) and magnesium-chelatase subunit ChlI-2 (Chli2) (Figure 6A), which are respectively down- and upregulated in atnp01. Elo3 is one of the six subunits of Arabidopsis Elongator Complex (AtELP) [73]. Atelp/elo mutants display pleiotropic phenotypes, including resistance to oxidative stress, hypersensitivity to abscisic acid, severely aberrant auxin phenotypes, altered cell cycle progression, abnormal root development, and disease susceptibility [74–76]. Recent studies have shown that AtELP3, together with AtELP2, regulates the kinetics of pathogen-induced transcriptome reprogramming [77]. CHLI isoforms in Arabidopsis are encoded by two genes: Chli1 (At4g18480) and Chli2 (At5g45930). Chli1 seems to be the major functional form since chlorophyll levels in chli1-null mutants are reduced to 10% to 17% of the wild-type level [78]. It has been hypothesized that Chli2 could support some chlorophyll biosynthesis in the complete absence of Chli1 [79]. At1g13880 (Elm2) is a member of the MYB class of transcription factors. Elm2 has only a limited role in mitochondrial fission, but it is a 54% paralog with ELM1 which has a fundamental function in mitochondrial fission [80]. Elm2 interacts with one differentially abundant upregulated protein, NifU-like protein 3 (Nfu3), which is involved [81] in the cluster assembly of chloroplastic Fe-S proteins (Figure 6A).

The *atnp*02 mutagenic event affected genes on chromosome 3, potentially At3g24330, encoding an O-glycosyl hydrolase localized in the endomembrane, and At3g24430 (Hcf101) encoding an ATP binding protein localized in the chloroplast. The mutagenic element lay within the At3g24400 pseudogene (AtPERK2), which possibly encodes a proline-rich extensin-like receptor kinase [14]. Hcf101 results were bound in the network with Nfu3, which in this case is downregulated (Figure 6B). The gene Perk2 interacts with the downregulated protein phosphatase 2C (At4g28400) (Figure 6B). This protein has already been described in Sections 3.1 and 3.3 and is the only one in common among the wt and the two mutants. The hypothesis is that the proteins related to these mutated genes were also the main targets in the signal cascade deriving from interaction with CdS QDs during the treatment.

**Figure 6.** Molecular interaction network of differentially abundant proteins. The mutated genes in the two mutant lines are shown in red in all the networks. A schematic above each network shows the structure of the mutated genes as in [14]. The segment in red and yellow is the insertion element (Ac), the segments in black are the mutated genes and Chr indicates the chromosomes on which these genes are positioned. The putative proteins that were found in the protein analysis and that connect in the network are indicated with blue rectangles. Networks for mutant *atnp*01 (**a**) and mutant *atnp*02 (**b**). Colored lines and dots indicate different types of interaction evidence (cyan, from curated databases; green, gene neighborhood; blue, gene co-occurrence; pink, experimentally determined; black, co-expression; light blue, protein homology).

#### *3.5. Comparison between 2D-PAGE, Pf2D, and Transcriptomics*

Marmiroli et al. [25] compared the proteomes of wt and the two mutants, treated with 80 mg/L CdS QDs, using a gel-less technique based on liquid IEF in the first dimension and HPLC in the second dimension, performed with a ProteomeLabTM PF2D (Beckman Coulter) [25]. The IEF gradient was pH 4–8.5 and the most water-soluble proteins were separated with this method. In our experiments we found four proteins in common with Marmiroli et al. [25]. For the wild type: ribulose bisphosphate carboxylase large chain (rbcL) (Rubisco large chain) [82], (downregulated in both cases); glutathione S-transferase DHAR3 (Dhar3) and pathogenesis-related protein 5 (At1g75040) (both proteins upregulated in all cases); for the mutant *atnp*02: heat shock 70 kDa protein 10 mitochondrial (Hsp70-10) (upregulated). For the mutant *atnp*01 there were no common proteins.

Plant pathogenesis-related (PR) proteins are expressed in response to pathogen attack, developmental processes and environmental stresses; some of these proteins are constitutively expressed [83]. Hsp70-10 was required for transport of secretory proteins from the Golgi complex [84]. Comparing our data with the transcriptomics results reported in Marmiroli et al. [14], we found three proteins in common: osmotin-like protein OSM34 (Osm34), pathogenesis-related protein 5 (At1g75040) and phosphatidylinositol/ phosphatidylcholine transfer protein SFH3 (Sfh3). OSM34 was always upregulated both in the transcriptomics and in the proteomics in the wt and in the two mutants. This protein is usually considered a response protein to osmotic stress [85]. Pathogenesis-related protein 5 was upregulated in the transcriptomics, and in the proteomics was also upregulated in both wt and mutant *atnp*01. This protein was in common with the gel-less proteomic study performed with PF2D (see above). For all lines Sfh3 was found downregulated in the transcriptomic, while in our study was upregulated in *atnp*01 and downregulated in wt

and in *atnp*02. These common proteins are mostly related to the oxidative stress response, which seems to be the driving stress arising from the interactions with CdS QDs [15].

Figure 7 shows the levels of correlation between proteomics (2D-PAGE)/transcriptomics, proteomics (PF2D)/transcriptomics, and proteomics (2D-PAGE)/proteomic (PF2D) markers. The figure was obtained by comparing 98 significant proteins obtained with 2D-PAGE with 88 selected proteins obtained with PF2D and the significant group of transcripts obtained with an Arabidopsis microarray platform. It is well known that the correlation between proteomics and transcriptomics is moderate to low [25,29], and in this case is also strongly biased by the difference between the higher number of transcripts and the lower number of proteins. Therefore, the proteins/transcripts considered as molecular markers showing this degree of correlation in the three comparisons are viewed as robust enough to be considered candidate "-omics" CdS QDs exposure markers.

**Figure 7.** Scatterplot matrix of the independent variables: transcriptome, proteome with 2D-PAGE, and proteome with PF2D. The three-dimensional scatterplot represents the correlation among gene expression, protein abundance with 2D-PAGE and protein abundance with Pf2D. The lines are represented in black for wt, in red for *atnp*01, and in green for *atnp*02. Transcriptomics data are taken from array analysis in Marmiroli et al. [14], and proteomic with PF2D data from Marmiroli et al. [25]. Each point represents a protein/transcript for the different plant lines using colors as above.

#### **4. Conclusions**

The majority of the differential abundance proteins in the wt were downregulated on exposure to CdS QDs and in Mapman bins annotated to processes like protein folding, biotic and abiotic stress responses, and protein degradation. Conversely, in both mutants, there was a balance in numbers between reprogrammed up- and downregulated proteins (Figure 2). Mapman bins for the mutant *atnp01* were protein degradation, abiotic stress and mitochondrial electron transport. These pathways are all typical of the responses to nanomaterials [86]. For the mutant *atnp02*, the main Mapman bins were glycolysis, mitochondrial electron transport, photosynthesis and hormone metabolism [15,86]. These results indicate that in *atnp02* the photosynthetic apparatus was impaired by treatment with CdS QDs, in addition to the other categories, especially hormone metabolism and mitochondrial electron transport chain, which were common to both mutants.

Network analysis showed that in the two mutants, the genes affected by the transposons are responsible for regulation of four proteins: *Nfu, Elo3, Chli2* and *PP2c*, involved respectively in chloroplast assembly, transcription elongation, chlorophyll biosynthesis and abiotic stress response. The reprogramming of these particular proteins demonstrates the importance of the chloroplast and of photosynthesis in the responses to CdS QDs of the mutants.

The authors are perfectly aware that there are more powerful proteomic tools available for plants (i.e., iTRAQ) which can give higher resolution and better quantification than 2D-PAGE. The purpose of this paper was to compare protein variations resulting from samples treated with CdS QDs vs. untreated controls, which certainly may benefit from powerful proteomic tools, but at the same time to probe these variations with the aid of a genetic tool, two CdS QDs-tolerant mutants. The use of the two mutants allowed refining the protein comparison from a large systematic and almost taxonomic level to a comparison in which a small proportion of proteins were objectively important. The two mutants consistently narrowed the proteomic range, allowing focus on those proteins specific for the tolerant phenotype rather than attention being dispersed on a plethora of proteins that were more aligned to networking effects. This approach is suitable not only following treatment with CdS QDs, but also for many other stressed conditions.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/2079 -4991/11//615/s1, Supplementary File 1: S.1. Supplementary methods related to nanoparticles synthesis and characterization. S.2. MapMan pathways identified under QDs treatment. Figure S1. HRTEM image of ligand-free QDs assembly and X-ray diffraction pattern. Figure S2. (A) ESEM image of the CdS QDs agglomerates. (B) X ray spectra corresponding to the red rectangle in figure S2A. Figure S3. 2D SDS-PAGE. Figure S4. Venn diagrams for common and non-common proteins to wt and *atnp01*, to wt and *atnp02*, to *atnp01* and *atnp02* (a) in the control and (b) in treatment conditions. Figure S5. Heat map of *A. thaliana* wt and mutant lines *atnp01* and *atnp02* not treated and treated with 80 mg L−<sup>1</sup> CdS QDs. Figure S6. Gene Ontology and enrichment analyses with fold enrichment = −log10 (Fisher's exact *p* value) for molecular function wt (a); cellular component wt (b); molecular function *atnp01* (c); cellular component *atnp01* (d); molecular function *atnp02* (e), cellular component *atnp02* (f). Figure S7. Cell function overview map after CdS QDs exposure. Cell functions associated with the proteomic changes affecting *Arabidopsis thaliana* after CdS QDs exposure in wt (a) *atnp01* (b) and in *atnp02* (c) using MapMan software. Figure S8. Biotic stress overview map after CdS QDs exposure. Stress response associated with the proteomic changes affecting A. thaliana after CdS QDs exposure in wt (a) *atnp01* (b) and in *atnp02* (c) using MapMan software. Table S2. MapMan BIN assignation and description of differentially expressed proteins in *Arabidopsis thaliana.* Supplementary File 2: Table S1. MALDI-TOF/TOF data associated with differentially expressed proteins identified by 2D-PAGE.

**Author Contributions:** N.M., V.G., M.M. organized the experimental setup and collaborated in writing the manuscript. V.G. performed the experiment, analyzed the data and helped in writing the manuscript. M.V. and A.Z. provided the CdS QDs and their physico-chemical characterization. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Publicly available datasets were analyzed in this study. This data can be found here: [doi:10.1021/es404958r; doi.org/10.3389/fpls.2015.01104].

**Acknowledgments:** The authors acknowledge the assistance of Gianluca Paredi (Department of Pharmacy, University of Parma) for MS analysis, of CIM (Parma, Italy) for access to MALDI-TOF-MS facilities. Martin Shepherd is acknowledged for helpful suggestions and language revision.

**Conflicts of Interest:** The authors declare no conflict of interest.
