Transcriptomic and proteomic profiling of maize embryos exposed to camptothecin

Background Camptothecin is a plant alkaloid that specifically binds topoisomerase I, inhibiting its activity and inducing double stranded breaks in DNA, activating the cell responses to DNA damage and, in response to severe treatments, triggering cell death. Results Comparative transcriptomic and proteomic analyses of maize embryos that had been exposed to camptothecin were conducted. Under the conditions used in this study, camptothecin did not induce extensive degradation in the genomic DNA but induced the transcription of genes involved in DNA repair and repressed genes involved in cell division. Camptothecin also affected the accumulation of several proteins involved in the stress response and induced the activity of certain calcium-dependent nucleases. We also detected changes in the expression and accumulation of different genes and proteins involved in post-translational regulatory processes. Conclusions This study identified several genes and proteins that participate in DNA damage responses in plants. Some of them may be involved in general responses to stress, but others are candidate genes for specific involvement in DNA repair. Our results open a number of new avenues for researching and improving plant resistance to DNA injury.


Background
Maintenance of genome stability is of critical importance for all organisms. Genomic DNA is continuously subject to many types of damage resulting from endogenous factors (production of reactive oxygen species, stalled replication forks, etc.) or the action of exogenous agents (radiation, naturally occurring radioisotopes, chemical mutagens such as heavy metals, etc.) [1]. Doublestrand DNA breaks (DSBs) are one of the most serious forms of DNA damage, potentially causing chromosomal translocations and rearrangements [2]. In response to DSBs, cells initiate complex signalling pathways that activate DNA repair, cell-cycle arrest, and eventually cell death [3]. DSBs repair is mediated by two basic mechanisms: homologous recombination (HR) and non-homologous end joining (NHEJ) [4]. In HR, an intact copy of the damaged region (a sister chromatid, for example) acts as a template to repair the break. In NHEJ, DSBs are simply rejoined largely independently of the DNA sequence. Bacteria and yeast usually employ HR whereas mammals and plants usually use NHEJ.
In addition to the direct repair of DNA breaks, additional responses are activated during DNA-damage stress. For example, DNA damage in plant cells usually induces the accumulation of signal transduction intermediates such as nitric oxide, ROS or ethylene [5,6] and produces changes in the cytosolic-free Ca 2+ [7]. It also induces cell cycle arrest, the inhibition of DNA and RNA synthesis, and a rapid protein turnover via the proteasome [8,9]. Additional reported effects are a reduction in the photosynthesis-related proteins [10], the accumulation of protective proteins such as pathogenesis-related protein-1 [11], the accumulation of protecting pigments [12], an increase in the expression of senescence-and cell death-associated genes [13] and the activation of different cellular detoxification mechanisms [14]. The regulation of all these responses is complex and involves different levels of regulation, including the modulation of transcriptional activity [15], post-transcriptional mechanisms (RNA processing, RNA silencing, etc.) [16][17][18] and post-translational modifications (protein phosphorylation, ubiquitination, SUMOylation, etc.) [19]. These processes are based on signal transduction initiated by sensor proteins that recognise the damage in the DNA and activate the transducers, which send the signal to the effector proteins [20]. The network of transcriptional, post-transcriptional and posttranslational modifications ensures temporally and spatially appropriate patterns of stress-responses.
DNA topoisomerase I (TOPI) regulates the topological state of DNA by cleaving and re-joining one DNA strand and allowing DNA relaxation [21]. TOPI activity is essential in dividing cells to release the torsion created by the progression of DNA replication forks. The presence of active TOPI is essential for embryo development in Drosophila and mouse [22]. In plants, TOPI plays a similar basic role and, for example, the disruption of the two TOPI encoding genes in Arabidopsis thaliana is lethal [23]. Camptothecin (CPT) is a plant alkaloid that specifically binds TOPI, stabilising the complexes formed between DNA and TOPI [24]. The collisions between the trapped TOPI-CPT complexes and the replication fork during DNA replication produce DSBs which induce DNA damage responses [25]. In consequence, actively dividing cells are much more sensitive to CPT than non-dividing cells, a property that has been exploited in the treatment of cancer [24]. However, non-dividing cells are also sensitive to CPT as collisions of the RNA polymerase machinery with the TOPI-CPT complexes, although less frequent, are also able to produce DSBs [26]. CPT-mediated TOPI-DNA complexes can be degraded via the 26S proteasome pathway so, at low CPT concentrations, cells can survive [27]. However, in actively dividing cells the high number of collisions may exceed the capacity of the cells to eliminate TOPI-DNA complexes and the DNA repair capability of the cells and, under these circumstances, cell death is initiated. CPT has a similar effect on TOPI in plant and animals. For example, CPT inhibits, in vitro, the activity of TOPI extracted from maize immature embryos [28], produces the abortion of shoots and roots in Arabidopsis [23], and induces cell death in tomato cell cultures [29].
In this study, we profiled proteins and genes whose expression is changed in immature maize embryos as a consequence of the DNA damage produced by CPT. Immature embryos contain a high proportion of actively dividing cells and, in consequence, are particularly sensitive to CPT. The combination of microarray and twodimensional gel electrophoresis protein analysis allowed us to identify molecular events that are regulated during DNA repair responses in plants at different levels: transcriptional, post-transcriptional, translational and posttranslational. We identified candidate genes and proteins which may be specifically involved in the DNA repair responses.

Results
Camptothecin induces DNA damage responses in maize immature embryos but not an extensive cell death process Maize caryopses were collected 15 days after pollination and their dissected embryos incubated, in the dark, in culture medium with or without 50 μM camptothecin (CPT). After 8 days of culture, the germination rates of treated and untreated embryos were not significantly different (24% ± 5 in control and 20% ± 6 in treated embryos) and their morphological characteristics were similar.
CPT is a DNA damaging agent that induces DNA repair responses [24], while ribonucleotide reductase (RNR) is an enzyme that provides dNTPs for DNA repair, with RNR genes being induced in response to DNA damage [30]. In order to check if, under our conditions, CPT is able to induce DNA repair responses in maize embryos, we used a maize gene ZmRNR2 probe encoding the ribonucleotide reductase, in northern blot hybridization ( Figure 1A and 1B). There was a high level of accumulation of the ZmRNR2 mRNA after 3 days of CPT treatment and, although reduced, the accumulation was maintained after 8 days of treatment (Figure 1A). The induction of ZmRNR2 was much higher in the embryo axis than in the scutellum ( Figure 1B).
Nucleases are involved in DNA damage responses [31] and in cell death [32]. In plants, cell death-related nucleases have been classified according to their cationic cofactors, as Ca 2+ or Zn 2+ -dependent [33]. The ability of CPT to induce nuclease activities in maize embryos was tested using in-gel nuclease activity assays ( Figure 1C  and 1D). An increase in the activity of a Ca 2+ -dependent nuclease of about 32 kDa was clearly evident after 3 days of CPT treatment using an assay buffer containing 1 mM CaCl 2 , being only slightly reduced after 8 days of treatment ( Figure 1C), and was higher in the embryo axis compared to scutellum ( Figure 1D). In contrast, no zinc-dependent nuclease activity was detected using 1, 2 or 5 mM Zn 2+ (results not shown).
The CPT-induced Ca 2+ -dependent nuclease could be involved in DNA repair but also in programmed cell death (PCD). PCD is usually characterised by internucleosomal genomic DNA fragmentation, producing, after gel electrophoresis, a characteristic DNA ladder pattern [34]. The results of electrophoresis of genomic DNA extracted from treated embryos was not significantly different to that observed with untreated embryos, showing a certain DNA ladder ( Figure 1E). The same analyses using DNA extracted separately from embryo axis and scutellum clearly show that the DNA ladder was only present in the scutellum sample ( Figure 1F). Degradation in genomic DNA extracted from scutellum has been previously observed in maize [34]. Cells in the scutellum close to the embryo axis undergo PCD as a normal part of seed development and this may explain the observed DNA ladder [35]. Exposure to 50 μM CPT did not, however, produce a significant change in the DNA integrity in the embryo axis or scutellum. This suggests that the CPT-induced Ca 2+ -dependent nuclease is involved in DNA repair and not in cell death.
In situ detection of fragmented DNA (TUNEL), as a sensitive technique to detect the initial steps of genomic DNA degradation, was used to analyse the effects of CPT on maize embryo DNA integrity ( Figure 2). In accordance with published data [35], untreated embryos only showed TUNEL positive nuclei in the scutellum, close to the embryo axis (Figures 2A and 2C). There was no increase in the number of positive nuclei in the scutellum of CPT treated embryos ( Figure 2B and 2D). The embryo axis of untreated embryos did not show any TUNEL staining ( Figure 2E). On the contrary, in CPT-treated embryos, some cells in the embryo axis showed TUNEL stained nuclei ( Figure 2F). However, the proportion of cells with stained nuclei was not high, which may explain why we did not observe extensive genomic DNA degradation in gel electrophoresis.
These results indicate that, under the conditions used here, CPT induced DNA repair responses in maize embryos but not an extensive cell death process. In-gel nuclease activity assay of total protein extracts (10 μg) of immature maize embryos treated with 50 μM CPT for three (E3D) and eight (E8D) days. The nuclease activity is detected as a non-stained halo in a polyacrylamide gel containing DNA stained with ethidium bromide. The deduced weight of the proteins with nuclease activity is indicated on the left (kDa). (D) Ingel nuclease activity assay of total protein extracts (10 μg) of dissected embryo axis (EA) and scutellum (SC) of immature maize embryos treated with 50 μM CPT for three days. The deduced weights of the proteins with nuclease activity are indicated on the left (kDa). (E) Integrity of nuclear DNA (4 μg) of immature maize embryos treated with 50 μM CPT for three (E3D) and eight (E8D) days, assayed by electrophoresis on 1.5% agarose gels. (F) Integrity of nuclear DNA (4 μg) of dissected embryo axis (EA) and scutellum (SC) of immature maize embryos treated with 50 μM CPT for three days, assayed by electrophoresis on 1.5% agarose gels.

Transcriptional responses to CPT-induced DNA damage
A global picture of the changes in gene expression produced during CPT treatment was obtained using the Affymetrix™ GeneChip Maize Genome Array. In this experiment, control and 3-day CPT-treated embryos were compared ( Figure 3). Ninety-three probe sets were found to have significantly increased or decreased signal in response to CPT, 39 up-regulated (Table 1) and 54 down-regulated ( Table 2). The probe set corresponding to the ZmRNR2 gene, previously used as a control for DNA damage response, was among the up-regulated genes. A quantitative real-time RT-PCR approach was used to validate the expression of 10 genes identified as differentially expressed in the microarray analysis, including 7 up-and 3 down-regulated genes ( Figure 4). Realtime PCR results were in very good agreement with the microarray data, although there were higher fold-changes using real time RT-PCR, which may be due to differences in the dynamic range and sensitivity of the two methods, as has been previously suggested [36].
The molecular roles of many of the altered genes remain unknown (31% of the up-regulated and 44% of the down-regulated). These genes may be involved in the control and/or execution of DNA damage responses ( Figure 5). DNA replication, recombination and repair (18%) and defense and stress responses (15%) were the two most abundant functional categories among the upregulated genes. Among down-regulated genes, the two most abundantly represented categories were signal transduction and gene expression (22%) and cell growth and division (17%). The functional category of DNA replication, recombination and repair was significantly more represented among the induced genes while the cell growth and division category was significantly more represented among the repressed genes ( Figure 5).
CPT treatment induced the expression of genes involved in DNA repair and DNA damage responses as, for example: -Two subunits of the ribonucleotide reductase: involved in the DNA repair processes [30].
-RAD51: encodes a protein required for meiosis and HR repair [37]. Maize mutants in two RAD51 maize genes are hypersensitive to radiation [38]. The Arabidopsis gene AtRAD51a is transcriptionally up-regulated by DSB-inducing agents and seems to be required for HR repair after bleomycin treatment [39].
-Rpa2: encodes a protein that is part of a heterotrimeric protein complex that specifically binds singlestranded DNA (ssDNA) and plays multiple roles in DNA metabolism, including DNA repair and recombination [40]. RPA genes are transcriptionally induced in Aspergillus nidulans exposed to CPT [41].
-TBPIP1: encodes a protein involved in chromosome pairing and segregation [42]. In humans, TBPIP1 enhances the strand exchange mediated by RAD51 [43]. In Arabidopsis, the TBPIP1 gene is transcriptionally induced by DNA damage [44].
-XRI-1: encodes a protein essential for meiosis and that plays a role during HR in Arabidopsis [45]. This gene is highly and rapidly transcriptionally induced by X-ray radiation and is also highly induced by other DSBs-inducer agents [44]. The encoded protein is probably part of the meiotic recombination complex MND1/ AHP2, which collaborates with RAD51 in the DNA strand invasion during recombination [46].
-Acetyltransferase, GNAT family protein: some yeast GNAT family members are involved in DSBs repair [47].
-Rph16: encodes a protein similar to RAD16 and is involved in the nucleotide excision repair of UV damage [48].
CPT treatment repressed the expression of genes involved in cell cycle, cell division and cell growth ( Table 2). For example: -Three cyclins: IaZm, IIZm and IIIZm.
-Shugosin-1: encodes a protein involved in the maintenance of centromeric cohesion of sister chromatids during meiosis and mitosis. Depletion of the human Sgo1 gene produces mitotic cell cycle arrest [49].
-TPX2: encodes a protein necessary for mitotic fuse formation in vertebrates [50]. The inhibition of the Arabidopsis TPX2 gene blocks mitosis [51].
-Knolle: encodes a syntaxin-like protein that acts during cytokinesis vesicle fusion and mediates cell-plate formation [52]. Knolle expression is repressed by gamma radiation in Arabidopsis [15].
-Patellin-5: patellins are involved in vesicle trafficking events. The Arabidopsis patellin PATL1 has been associated with the formation of the cell-plate during cytokinesis [53].
-Microtubule-associated protein RP/EB family member 3: encodes a protein that binds to the end of the microtubules and is important in maintaining the structure of the mitotic spindle [55].
-Growth regulating factor 8-like: encodes a protein involved in leaf and cotyledon growth in Arabidopsis [56].  -Rough sheath1: encodes a protein involved in cell differentiation [57].
-Frizzy-like protein/WD-repeat cell cycle regulatory protein: encodes a protein similar to the tomato CCS52B that probably is involved in cell-cycle control during mitosis [58].

Alterations in maize embryo proteome in response to CPT
Equal amounts of total protein extracted from control and from CPT-treated maize embryos were fractionated using 2-D gel electrophoresis ( Figure 6A and 6B). At least three-fold increase/decrease and t-test p < 0.05 were used as the criteria to select differentially accumulated polypeptides. In response to CPT treatment, 455 spots showed quantitative or qualitative (presence/ absence) variations between the two gels, with the intensity decreasing in 169 and increasing in 286. Some examples of up-or down-accumulated spots are shown in figure 6C. Forty-three of the spots with significant differential expression on gels were chosen for identification by MS/MS mass spectrometry. Interpretable MS/MS spectra were obtained for 31 spots. The location of these in the gels is shown in Figures 6A and  6B.
The identified proteins belong to a variety of functional categories (Table 3). For example, CPT alters the accumulation of two enzymes involved in glycolytic metabolism: glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and triosephosphate isomerase 1. Interestingly, in human neuronal cells, CPT also produces changes in the accumulation of GAPDH [59]. In plants, the accumulation of both proteins has been described in response to different types of stress [60][61][62][63][64].
Some of the identified proteins are involved in antioxidant responses. Antioxidant activity protects against ROS accumulation, which can be produced by a variety of stresses, including DNA damage [65]. In mammals, CPT induces the accumulation of antioxidant enzymes in the nucleus [66]. The accumulation of two proteins involved in pathogenesis responses, PR1 and Betv1, was observed in response to CPT. They are also induced by      abiotic stresses such as heavy metals [67] and UV radiation [68]. The accumulation of at least two 26s proteasome regulatory subunits is altered in response to CPT, one increased and the other reduced. Interestingly, CPT-TOPI-DNA complexes may be degraded by the ubiquitin-dependent pathway [27].
We observed changes in the accumulation of several proteins involved in RNA metabolism or RNA binding proteins: -RraA: an RNaseE inhibitor which may be involved in the degradosome complex in E. coli [69,70]. Regulation of RraA by DNA damage stress could be related to changes in the regulation of RNA homeostasis.
-DEAD-family RNA helicase: DEAD-RNA helicases act in RNA metabolism promoting either RNA synthesis or decay [71]. Some have been associated with abiotic stress [72].
-Glycine-rich RNA-binding protein 8: some glycinerich RBPs in Arabidopsis (GR-RBPs) are significantly induced by cold, drought and salinity, whereas others are repressed by other sources of stresses [73].
We also observed an increase in the accumulation of two spots corresponding to eukaryotic elongation factors. Experiments in yeast and mammals demonstrate that translation initiation factor 5A (eIF5A) is actually involved in mRNA nucleus-cytoplasm export and not translation, specifically regulating genes involved in cell growth and proliferation, and in cell death [75]. In Arabidopsis, AteIF5A/AtFBR12 (At1g26630) promotes PCD associated with the hypersensitive pathogen response [76], and AteIEF5A-1 (At1g13950) has been associated with PCD during xylogenesis [77]. Thus, regulation of eiF5A by CPT suggests it is involved in cell cycle and PCD regulation.

Lack of correlation between CPT-induced changes in protein abundance and changes in mRNA accumulation
This study provided data on the most differentially expressed genes in control and CPT-treated embryos, and the most differentially accumulated proteins, allowing us to compare the datasets. The genes encoding 24 of the 31 identified proteins are represented in the microarray, but there was no significant change in expression in response to CPT (Table 4). This was confirmed by northern blot hybridizations using probes corresponding to nine of these genes, with no significant differences in the hybridization intensities observed (Figure 7).

Discussion
Our aim was to identify new elements involved in cellular responses to genomic damage in plants, using CPT as a toxic agent and applying transcriptomic and proteomic approaches to identify the genes, proteins and cellular mechanisms involved. We identified a series of genes and proteins whose expression/accumulation significantly change in response to CPT, although the identified genes do not correspond to the identified proteins. These differences may be a consequence of the different sensitivity of the methods. Moreover, the level of protein accumulation does not necessarily agree with the level of mRNA expression. This incongruent expression between mRNAs and proteins has been observed by other groups, in other species and experimental conditions [78][79][80] and is most likely a result of the biology of gene expression which includes various levels of regulation during protein synthesis: post-transcriptional, translational, and post-translational. Thus, integrated analysis of both mRNAs and proteins is crucial to gain further insights into complex biological systems.
The basic mechanism of action for CPT has been well-studied and characterised in animal cells [24]. CPT generates replication-mediated DSBs in DNA which in turn induce DNA repair, cell cycle arrest and, under certain circumstances, cell death. Under our conditions, CPT did not induce extensive cell death in maize embryos, as demonstrated by TUNEL staining which only appeared in some cells in the embryo axis after CPT-treatment. At the developmental stage analyzed here, cells in the scutellum divide at a very limited rate, but cells in the embryo axis divide rapidly. This difference may explain the higher sensitivity to CPT of the cells in the embryo axis.
Two basic mechanisms of DSBs DNA repair have been described: homologous recombination and non-homologous end joining [4]. Our transcriptomic analysis identified the induction of some genes already known to be involved in DNA repair. Interestingly, most of them are involved in the HR repair pathway, suggesting that this is the main mechanism for DSBs repair in maize embryos, at least in response to CPT. CPT also produces an increase of a 32 kDa calcium-dependent nuclease activity. However, this nuclease is unlikely to be involved in the extensive fragmentation of the genomic DNA observed in different cell death processes as extensive DNA fragmentation was not observed. Nucleases are also involved in most DNA repair mechanisms, including HR [81]. These data suggests that the 32 kDa nuclease activity observed may be involved in the DNA repair process.
CPT induces reversible or permanent cell-cycle arrest in G2-M phase in human and other cells [82] and produces major alterations in the expression of cell-cycle regulatory genes [83]. We found that CPT reduces the expression of several mitosis-related genes. In addition, we observed a reduction in the accumulation of the histone H2B involved in the structure of chromatin, and changes in the accumulation of two eukaryotic translation initiation factors which seem to also be involved in the cell-cycle process [84]. These results suggest that, in maize embryos, one of the cellular responses to CPT is the arrest of cell division.
In addition to more specific processes, DNA damage induces general stress mechanisms in maize embryos. For example, we observed changes in the expression and accumulation of proteins involved in ROS processing (glutathione S-transferase, Class III peroxidase precursor, chloroplast Cu/Zn superoxide dismutase, cytosolic ascorbate peroxidase), enzymes involved in glycolic metabolism (glyceraldehyde-3-phosphate dehydrogenase, triosephosphate isomerase 1) and in pathogen responses (pathogenesis-related protein1 and Bet v I allergen). Pathogen resistance is increased after DNA damage induction, indicating a cross-link in DNA damage (and maybe other abiotic stresses) and defense responses [85].
An increasing number of studies combining proteomics and transcriptomics clearly demonstrate that mRNA and protein accumulation are not always correlated [86][87][88]. For instance, in yeast 73% of the variance in protein abundance is explained by the translation mechanism and only 27% due to variations in mRNA concentration [89]. Protein abundance is influenced by several factors at the post-transcriptional, translational, and post-translational levels. For example, there is a time lag between transcription and translation in which introns are excised and the transcripts are moved from the nucleus to the cytoplasm, and translation rates may be influenced by ribosome, tRNA and amino acids availability, codon usage or accessory protein binding association [90]. In addition, protein abundance is also influenced by post-translational processes such as glycosylation, phosphorylation and proteolytic processing.
Our proteome analysis indicated differences in the abundance (up and down) of the encoded proteins of 24 genes whose mRNA levels do not significantly change in response to CPT (Table 4). It is possible that CPT induces the transcription of some genes only during the first hours of treatment, and after three days of treatment the mRNA levels are similar to the control but the abundance of the encoded protein is higher. Differences in the translation rate may also explain the lack of correlation. In animals, post-transcriptional regulation of gene expression during the stress response means specific stress-induced transcripts receive the highest transcriptional priority [91]. Interestingly, some of the spots identified in the proteomic analysis correspond to proteins associated with RNA metabolism and RNA binding proteins, and may be involved in the regulation of mRNA translation.  Table 3. (C) Portions of 2-D gels showing spots (arrows) that were differentially abundant between untreated (CON) and treated (CPT) embryos.  Many post-translational processes affect the position of a protein in 2D gels such that the protein appears as differentially accumulated in a proteomic analysis. We have identified changes in genes and proteins involved in protein modification and post-translational regulation. For example, the accumulation of at least two 26s proteasome regulatory subunits is altered in response to CPT and the expression of the proteasome inhibitor-like protein PI31 is increased. Ubiquitin/proteasomemediated protein degradation plays a central role in the regulation of several aspects of plant development and stress responses [92] and our data indicate that it may also be involved in regulating DNA damage responses. In fact, there are evidences that CPT-TOPI-DNA complexes may be degraded by the ubiquitin-dependent pathway in mammals and yeast [27,93]. Our data suggest that a similar situation may occur in plants. Moreover, the expression of embryonic flower 2 is repressed, a gene encoding a protein homologous to Drosophila Polycomb genes which mediate the epigenetic control of homeotic gene expression [94].
The role of several of the genes identified in the transcriptomic analysis is unknown. These genes may play a role in DNA damage detection and repair mechanisms, especially those genes that are only induced in response to genomic damage and not in response to other types of stress. Unfortunately the data currently available in maize does not allow us to determine which of them are specifically induced by DNA damage, but many of the maize identified genes have clear homologues in Arabidopsis (Table 1 and 2). Microarray analyses in Arabidopsis have been used to study the effects of several abiotic stresses, including two DNA damage agents, bleomycin [4] and gamma radiation [15]. Examining microarray databases [95] we identified eight Arabidopsis genes homologous to maize CPT-induced genes and exclusively induced by DNA damage: At5g02220, of unknown function; At1g13330, encoding TBP-1 tat binding protein; At5g48720, encoding an X-ray induced gene required for post-meiotic stages of pollen development and for male and female meiosis; At3g27060 and At2g21790, encoding the ribonucleotide reductase (RNR) small and large subunit, respectively; At5g20850, encoding AtRAD51; and two genes, At5g18270 and At3g04060, encoding NAC transcriptions factors. NAC proteins constitute one of the largest families of plantspecific transcription factors, and the family is present in a wide range of land plants [96]. These two NAC

Conclusions
The integration of microarray and proteomic analyses provides new data on DNA damage responses in plants. This is a complex process involving DNA repair and arrest of cell-cycle, but also general stress responses. Post-translational processing and the regulation of mRNA translation seem to have an important role in DNA damage responses.

Plant material and treatments
Maize (Zea mays L cv W64A pure inbred line) was grown under controlled conditions (16 h light, 28°C). Immature embryos (15 days after pollination) were extracted in sterile conditions and placed on MS plates (4.4% (w/v) Murashige and Skoog medium, 0.8% (w/v) Gelrite) supplemented (or not) with 50 μM camptothecin (Sigma-Aldrich) and maintained in a growth chamber at 26°C in darkness.

Histological analysis
Embryos were collected, fixed in ethanol-formaldehydeacetic acid (80:3.5:5) for 1 h at room temperature, followed by 1 week at 4°C, and then stored in 70% ethanol at 4°C. Fixed samples were embedded in paraplast, de-waxed with Histo-Clear II (National Diagnostic, UK), re-hydrated in an ethanol series and equilibrated in 0.02 M citric acid-0.16 M Na 2 HPO 4 , pH 7.0. TUNEL assays were done using the In Situ Cell Death Detection kit (Roche) according to the supplier's protocol for difficult tissues. In negative controls, the TdT enzyme was omitted, and the positive controls were treated with DNase I for 10 min. Experiments were repeated three times.

RNA extraction and quantification
Total RNA was isolated from frozen samples using the lithium chloride method. DNase digestion of contaminating DNA in the RNA samples was done using RNase-Free DNaseI. Final RNA purification was performed using the RNeasy Mini Kit (Qiagen) according to standard protocols. RNA was quantified with a Nano-Drop ND-100 spectrophotometer (NanoDrop Technologies). RNA quality was assessed with a 2100 Bioanalyzer from Agilent Technologies.

Affymetrix GeneChip hybridization/Microarray analysis
Gene expression was analyzed using the Affymetrix GeneChip ® Maize Genome Array, which contains probe sets to interrogate 13,339 genes, performing four independent biological replicates. cDNA synthesis, probe labeling, array hybridization and data analysis were as described by Bannenberg and col. [97], in the Genomics Service of the Centro Nacional de Biotecnologia (CNB-CSIC, Madrid). Raw data and normalised data were deposited at the ArrayExpress data library (http://www. ebi.ac.uk/arrayexpress/) under accession number E-MEXP-2702. Differential expression was considered following the p < 0.05 and 2.0 fold change as the criteria of significance. Functional categories of the genes were determined based on Gene Ontology data. We used the Fisher's Exact Test (p ≤ 0.05) and ANOVA (p ≤ 0.01) to determine the significant differences in the functional categories among up-and down-regulated genes.

Real time quantitative RT-PCR
To validate the expression changes found in the microarray experiments, transcript levels of the ten selected genes were quantified by the ABI Prism 7700 (Applied Biosystems, Foster City, CA, U.S.A) as described by Mascarell-Creus and col. [98]. The oligonucleotides chosen to amplify the selected genes were designed using the Primer Express Software (Applied Biosystems) and are listed in table 5. The actin gene was used as the internal control.
In-gel digestion of proteins and MS and MS/MS spectra Proteins were in-gel digested with trypsin and tryptic peptides were extracted and analyzed by MALDI-TOF/ MS (4700 Proteomics Analyzer, Applied Biosystems) or LC-ESI-QTOF (Q-TOF Global, Micromass-Waters) mass spectrometers in the Proteomics Platform (PCB) of the University of Barcelona as previously described [100].   The oligonucleotides were designed using the Primer Express software (Applied Biosystems).