Artificial Autopolyploidization Modifies the Tricarboxylic Acid Cycle and GABA Shunt in Arabidopsis thaliana Col-0

Autopolyploidy is a process whereby the chromosome set is multiplied and it is a common phenomenon in angiosperms. Autopolyploidy is thought to be an important evolutionary force that has led to the formation of new plant species. Despite its relevance, the consequences of autopolyploidy in plant metabolism are poorly understood. This study compares the metabolic profiles of natural diploids and artificial autotetraploids of Arabidopsis thaliana Col-0. Different physiological parameters are compared between diploids and autotetraploids using nuclear magnetic resonance (NMR), elemental analysis (carbon:nitrogen balance) and quantitative real-time PCR (qRT-PCR). The main difference between diploid and autotetraploid A. thaliana Col-0 is observed in the concentration of metabolites related to the tricarboxylic acid cycle (TCA) and γ-amino butyric acid (GABA) shunt, as shown by multivariate statistical analysis of NMR spectra. qRT-PCR shows that genes related to the TCA and GABA shunt are also differentially expressed between diploids and autotetraploids following similar trends as their corresponding metabolites. Solid evidence is presented to demonstrate that autopolyploidy influences core plant metabolic processes.


Results
Arabidopsis thaliana Col-0 diploids and autotetraploids analyzed at two ages (14-and 18-days-old) showed the same trends independent of their age. Thus analytical chemical and molecular biological results are shown only for 14-day-old plants. Colchicine-induced autopolyploidization of A. thaliana diploids was confirmed by the doubling in the amount of DNA for every peak detected in flow cytometry (Fig. S1). Multiple peaks are detected as a consequence of endoreduplication. Autotetraploids accumulated more biomass than diploids (Fig. 1). Diploid and autotetraploid A. thaliana above-ground tissue had a strongly different 1 H NMR metabolic profile as evidenced by the formation of separated clusters for each ploidy on the PCA score plot (Fig. 2). The loading plot showed that a group of highly variable bins aligned along PC1 (Fig. 3). 1D-STOCSY using 2.64 as driver peak, the most variable bin according to the loading plot, showed that bin 2.64 had a strong positive correlation with bins  Metabolic profiles of 14-day-old diploids (blue circles) and autotetraploids (red circles) are represented. Each dot represents the metabolic profile of a single plant. Each ploidy had a clearly distinct NMR-based metabolic profile as no overlap is observed between diploids and autotetraploids.  ). These signals matched with citrate in our database. Bin 2.64 was also positively correlated with bins 1.90, 2.28 and 3.00. These signals did not show cross peaks in the 1 H-13 C HSQC spectrum but they matched well with signals in our 1 H NMR database for γ -aminobutyric acid. Additional strong positive correlation was observed between bin 2.64 and bin 2.38, signal that corresponds to succinate in our database. Bin 2.64 correlated too with bins 2.34 and 3.76, and these signals showed cross peaks with 13 C signals at 36.25 and 57.05 ppm corresponding in our database with glutamate. Other bins showing high variability according to the loadings plot were bin 6.50, which showed a cross peak at 138.20 ( 13 C) that corresponds to fumarate in our database. All the identified metabolites participate in the tricarboxylic acid and GABA pathways. Other bins showing high variability in the loadings plot correspond to sugars. In the 1D-STOCSY bin 2.64 was not correlated with bin 6.50 but sugar signals were negatively correlated.
Gene expression levels showed comparable trends independent of the normalizing gene used, UBQ10 or ACT2. qRT-PCR analysis did not reveal any statistically significant difference for the expression of any of the studied genes between ploidies (Fig. 5a, Fig. S2). However, the mean values for most of the genes were higher in diploids, a tendency also observed for the 1 H NMR signal intensities of the corresponding metabolites (Fig. 5b). No difference was detected in the carbon:nitrogen ratios between ploidies (Fig. 6).

Discussion
Gene duplication is a powerful evolutionary force as it allows genetic novelty resulting from gene neofunctionalization 21,22 . If a gene is duplicated, then at least one of the two genes can be mutationally altered and survive purifying selection (because the mutant gene is covered) 23 . Autopolyploidization represents the most extreme case of  14-day-old diploid and autotetraploid 1 H NMR spectra were used for correlation. Driver peak: 2.64 (see experimental procedures). All compounds showing statistical correlation (signals displayed in red gradient) participate in the tricarboxylic acid cycle -TCA-(malate (a), citrate (b), and succinate (d)) or carbon:nitrogen balance (γ -aminobutyric acid (c), and glutamate (e)). gene duplication as all the genes in an organism are duplicated (whole-genome duplication) and in evolutionary time it can lead to speciation through the mechanism of reciprocal gene loss 24 . Recently formed autopolyploids are poised with immediate challenges connected with meiosis, e.g., restoring chromosome pairing and the establishment of a breeding population 25 . For example, Arabidopsis arenosa (a close relative of A. thaliana) is known to naturally exist as diploid and autotetraploid cytotypes and genome-wide data showed how natural selection has acted on meoisis genes 26 . Apart from these cytological challenges, neopolyploids also have to cope with the challenges imposed by the environment. Differences in the anatomy and in the timing of events in the life cycle could affect competition, environmental tolerances, and reproductive isolation of polyploids 27 . A well-known effect of autopolyploidization is an increase in cell and organ size 27,28 , as reflected in the bigger size of the autotetrapolyploid Arabidopsis thaliana Col-0 used in this study (Fig. 1).
Another crucial trait that deeply influences plant ecology is plant metabolism [29][30][31] . It is in this particular aspect of plant autopolyploidy that the present study pioneers. It presents the first evidence on how autopolyploidization modifies plant metabolism. Metabolites are the final expression of genes and about a century ago the formulation of the gene balance hypothesis implied that 'gene products [metabolites] interact with set stoichiometries and there is a tendency for purifying selection to not recognize whole-genome duplications because all genes in the genome change dosage contemporaneously, so balances are unchanged' 23,32 . The NMR results in this study contradict this notion as the variation in metabolite concentration is compound specific (Fig. 5b) and thus stoichiometries for the same gene products are different in diploids and autotetraploids. Moreover, variation in transcript levels is also gene product specific (Fig. 5a, Fig. S2). Previous reports show that autotetraploid A. thaliana Col-0 have an increased number of chloroplasts per cell as diploids 33 but it is not known if the higher chloroplast  density modifies photosynthetic rates in autotetraploids. Chloroplast and mitochondrial metabolisms are connected via metabolites of the tricarboxylic acid cycle -TCA- 34 . TCA metabolites identified in this study vary between diploids and autotetraploids (Figs 2 and 3). These metabolites also participate in cell guard metabolism and regulators of stomatal opening 35 . This opens the possibility that differences might exist between diploids and autotetraploids in terms of gas exchange, which in turn might be connected to different photosynthetic rates produced by the different chloroplast density. Among all the TCA metabolites varying between ploidies, malate displays the largest difference between diploids and autotetraploids. Functional roles for malate in plants are quite diverse including, but not limited to: respiration and energy generation, photosynthesis (both C3 and C4), fatty acid oxidation, lignin biosynthesis, nitrogen fixation and amino acid biosynthesis, ion balance, uptake of phosphorus and iron, aluminum tolerance and pulvinal and stomatal function 36 . Root excretion of malate has been associated with aluminum tolerance, phosphate nutrition and the establishment of microbial communities. A. thaliana Col-0 autotetraploids also have a lower concentration of glutamate and γ -amino butyric acid (GABA), both amino acids known to participate in the maintenance of carbon:nitrogen (C:N) balance [37][38][39] . The results of the present study show, however, that the differences in concentration for glutamate and GABA between diploids and autotetraploids do not affect the C:N balance regardless of the ploidy (Fig. 6). Glutamate and GABA are connected to the TCA cycle via a short pathway composed of three enzymes called the GABA shunt because it bypasses two steps of the TCA cycle. Remarkably, statistical correlations of the NMR spectra show that both TCA and GABA metabolites vary coordinately within plants (Fig. 4) suggesting that entire metabolic pathways are down regulated in the autotetraploids.
Interestingly, diploid and autotetraploid populations of A. arenosa are mostly geographically separated 40 . Similar geographic segregations have been observed in other multiploid plant species [41][42][43] . This allopatric distribution can be the result of reproductive incompatibilities between ploidies and has profound consequences for the ecology of plants. In the specific case of plant-insect interactions, differences have been detected in the insect species composition and extent of herbivory and pollination experienced by diploid and autotetraploid plants [44][45][46][47] . Unfortunately, these studies did not analyze volatile organic or defensive compounds, molecules known to regulate pollination and herbivory. The precursors of these specialized metabolites are produced by core metabolic pathways like the ones shown in this study to be modified by autopolyploidization in Arabidopsis thaliana. This fact opens the possibility that concentrations of specialized metabolites could also be different in individuals of different ploidy levels in multiploid plant species.
In summary, the metabolic differences produced by autopolyplody can have profound effects for the development and ecological interactions of plant neopolyploids. With at least 24 ancient autopolyploidization events identified to have occurred during the evolution of vascular plants 48 it is completely relevant to have a clearer picture of how this mutation can affect the fitness of the newly formed autopolyploids. The prevalent view is that autopolyploidy is a common and important element of plant diversity. This study suggests how divergent ecological interactions might arise from the physiological differences existing in neoautotetraploids and the ecological differences may in turn lead to differential evolutionary paths for autopolyploid plants.

Materials and Methods
Plant material. Arabidopsis thaliana Col-0 diploid plants were treated with colchicine as previously described to obtain autotetraploid seeds 49 . Diploid and autotetraploid seeds were sown in vermiculite and plants were kept at 25 °C and 50% relative humidity with 24 h light (75 μ mol m −2 s −1 ) and plenty of water. Plants were harvested 14 and 18 days after sowing. For each plant the whole above ground tissue was collected. Fresh tissue was used for flow cytometry. For all other analyses harvested plants were immediately flash-frozen with liquid nitrogen. Water was removed with a freeze-dryer and samples were immediately processed for chemical and molecular biological analyses. NMR spectra acquisition and processing parameters. Spectra were recorded using an Avance II 700 Bruker spectrometer equipped with a 5 mm inverse cryoprobe operating at 700.153 MHz for 1 H and 176.061 MHz for 13 C. Acquisition temperature was 298 K. 1 H spectra were acquired using a water suppression pulse program with the following conditions: 0.29 Hz point −1 , acquisition time = 1.67 s, relaxation delay = 2.5 s and 90° pulse width = 10 μ s. 128 transients with 4 dummies were recorded per spectrum. FIDs were Fourier-transformed using no line broadening (LB = 0.0 Hz). The resulting spectra were manually phased and baseline-corrected and calibrated to the internal standard (DSS-d6). 1 H-13 C HSQC spectra were acquired using the Bruker pulse program hsqcetgp for echo/antiecho gradient with the following conditions: relaxation delay = 1.5 s and 26,410 Hz spectral width in f1 and 9,803 Hz in f2. Qsine (SSB = 2.0) was used for the window function. 16 dummies and 48 transients were collected. FID was Fourier-transformed and the resulting spectrum was manually phased and baseline corrected and calibrated using the internal standard (DSS-d6).
Statistical analysis of 1 H NMR. Phased and baseline-corrected 1 H NMR spectra were binned from 8.0 to 0.5 ppm with a bin width of 0.02 ppm using the software automics 50 . Binning was performed applying the total scaling function of automics and the region from 5.0 to 4.6 ppm (HDO signal) was not binned. The matrix with the data of all the binned spectra (355 bins per spectrum) was exported to the software R and processed with the package muma for NMR metabolomics 51 . Principal component analysis (PCA) and 1D-statistical total correlation spectroscopy (1D-STOCSY) analysis were performed according to muma's manual and applying the following parameters: scaling = "pareto", imput = "mean", normalize = TRUE. Batch spectral integration of selected regions of interest (ROI) was performed on 1 H NMR spectra using rNMR 52 . The next ROIs (expressed in ppm) were used: 2.570-2.547 (citrate), 2.680-2.635 (malate), 6.513-6.500 (fumarate), 2.398-2.389 (succinate), 3.771-3.741 (glutamate) and 3.020-2.987 (γ -aminobutyric acid). t-tests were performed on single ROIs in order to identify significant signal intensity differences between ploidies. 1D-STOCSY was applied to the binned data in order to generate a pseudo-NMR spectrum displaying the correlation among the intensities of various peaks across the whole dataset, thus taking advantage of the multi-colinearity of the intensity variables in a set of spectra 53 . Candidate compounds based on 2D NMR. In order to increase signal intensity 15 diploid and 15 autotetraploid 14-day-old A. thaliana extracts were mixed, solvent freeze-dried and sample reconstituted in D 2 O. Coordinates for the resulting 1 H-13 C HSQC signals were automatically picked up in Topspin (Bruker). The resulting list of 32 1 H-13 C cross peaks was imported into the software spinassign [http://prime.psc.riken.jp] for signal matching against a database of standard compounds.
qRT-PCR analysis. Quantitative real-time PCR was performed on the freeze-dried above-ground tissue.
RNA from 2 plants was pooled to form 1 sample, and 4 samples per ploidy were analyzed (8 plants in total per ploidy). Conditions for mRNA extraction and gene expression analyses were as previously described 54 . Actin 2 and ubiquitin 10 mRNA concentrations were determined for each sample to normalize expression of individual genes. Oligonucleotides used are shown in the supporting information section (Table S1). t-tests were performed on single genes in order to identify significant expression level differences between ploidies.
Elemental analysis. Total carbon and nitrogen were determined on freeze-dried above-ground tissue using a vario MICRO cube elemental analyzer (Elementar). 10 diploid and 10 autotetraploid plants were used. Plant material was grounded and 1 mg of powder per plant was mixed with 1 mg WO 3 , placed on a Sn capsule and pressed to form a homogeneous pellet. The temperatures of the combustion and reduction tubes were set at 950 °C and 550 °C, respectively. Sulfanilamide was used according to vendor's instructions as data calibrator.