The role of CAX1 and CAX3 in elemental distribution and abundance in Arabidopsis seed.

The ability to alter nutrient partitioning within plants cells is poorly understood. In Arabidopsis thaliana , a family of endomembrane cation exchangers (CAXs) transports 3 Ca 2+ and other cations. However, experiments have not focused on how the distribution 4 or partitioning of Ca and other elements within seeds are altered by perturbed CAX 5 activity. Here we investigate Ca distribution and abundance in Arabidopsis seed from 6 cax1 and cax3 loss of function lines and lines expressing deregulated CAX1 using 7 synchrotron x-ray fluorescence microscopy. We conducted 7 – 10 µm resolution in vivo 8 x-ray microtomography on dry mature seed, and 0.2 µm resolution x-ray microscopy on 9 embryos from lines over-expressing deregulated CAX1 (35S-sCAX1) and cax1cax3 10 double mutants only. Tomograms showed an increased concentration of Ca in both the 11 seed coat and embryo in cax1 , cax3 and cax1cax3 lines compared to wild type. High 12 resolution elemental images in the mutants showed that perturbed CAX activity altered 13 Ca partitioning within cells, reducing Ca partitioning into organelles and/or increasing Ca 14 in the cytosol, and abolishing tissue-level Ca gradients. In comparison with traditional 15 volume-averaged metal analysis, which confirmed subtle changes in seed elemental composition, the collection of spatially-resolved data at varying resolutions provides insight into the impact of altered CAX activity on seed metal distribution and indicates a 18 cell type-specific function of CAX1 and CAX3 in partitioning Ca into organelles. This 19 work highlights a powerful technology for inferring transport function and quantifying 20 nutrient changes.

In this study we used two synchrotron microprobes with different spatial 3 resolutions to collect elemental images from seed of lines with altered CAX expression. 4 Images suggested that overexpression of a CAX1 with a N-terminal truncation in the 5 regulatory region (sCAX1), caused a disruption of selective Ca accumulation by cell 6 types; rendering all layers equally Ca-rich, and that deletion of both CAX1 and CAX3 Columbia wild type ( Figure 1B). For CAX1, the transcript was at low levels in all the 1 cDNA samples and was undetectable after 30 cycles of amplification (data not shown).
2 Even after 35 cycles, CAX1 bands were weak (data not shown). At 35 cycles, non-3 specific amplification also becomes a factor that severely decreases the accuracy the 4 measurement of very low levels of transcripts of interest. Thus, no conclusion was drawn 5 regarding the relative levels of CAX1 transcripts in cax3-1, Columbia wild type and 35S-6 sCAX1. 7 8 SXRF microtomography 9 We conducted SXRF microtomography to analyze the impact of CAX1 and 1 0 CAX3 disruption on the spatial distribution of elements within mature seeds ( Figure 2).

1
Microtomography of intact seed showed no distributional anomalies at the tissue level, of 1 2 either macronutrients (K and Ca) or micronutrients (Fe, Mn and Zn), nor was elemental 1 3 allocation between seed coat and embryo disrupted (Supplemental Figure 1). There were 1 4 differences in elemental abundances between lines, with higher K, Ca and Zn in cax1 and 1 5 cax3, and lowest abundances of these elements in 35S-sCAX1.

6
Volume-averaged seed analysis via ICP-MS (Table 1)  were highest in cax1-1 and cax3-1, Mn was lowest in 35s-sCAX1 whereas Zn was 1 9 highest in cax1-1 and cax3-1 and lowest in 35s-sCAX1. However, absolute values 2 0 between these two datasets differ, and the extent to which K, Ca and Zn differ appears is 2 1 less pronounced in volume-averaged data. We offer several explanations for this. First, 2 2 volume-averaged data is the average elemental concentration of a large number of seeds 2 3 (there are approximately 4000-5000 seeds per 50 mg aliquot), and considers the whole 2 4 seed volume. This is compared with the elemental composition of a 10 µm-thick slice 2 5 through a single Arabidopsis seed. In volume-averaged data, averaging over a large 2 6 number of seeds will reduce between-seed elemental variability, whereas in spatially-2 7 resolved data, the inherently low replicate number enhances variability. Additionally, 2 8 spatially resolved data is presented as maximum pixel abundance, indicating only the 2 9 upper limit of the data, rather than the average. For this reason, we have conducted region of interest (ROI) analysis on identified 1 tissue layers to derive their mean elemental abundances. These tissues consisted of seed 2 coat (Table 2a) and embryo (Table 2b), showing macronutrients K and Ca, and the 3 micronutrients Fe, Mn, Cu and Zn. Table 2a shows that Ca abundances in the seed coat of 4 cax1-1, cax3-1 and cax1cax3 are higher that wild type, and lower in 35S-sCAX1. The 5 high standard deviations of Fe and Mn in the embryo are due to their discrete distribution 6 in specific tissue layers, which were analyzed separately (Table 2c)  We conducted higher resolution (sub-micron) spectroscopy to fully investigate the 1 7 elemental distribution of subcellular compartments. High-resolution SXRF mapping (0.2 1 8 µm 2 beam) of the Ca distribution of whole embryo thick sections of wild type, 35s-1 9 sCAX1 and cax1cax3 mutants respectively are shown in Figure 3A, D and G. We chose 2 0 the double knockout and lines expressing deregulated CAX1 (35s-sCAX1) with the 2 1 assumption that elemental differences would be more pronounced than in the single 2 2 mutants (Conn et al., 2011).

3
Calcium was present in all cells, both enclosed within numerous organelles at 2 4 higher pixel abundances, and as a component of the cell walls at several orders of 2 5 magnitude lower abundance (arrow, Figure 3C). The Ca abundance range typically 2 6 spanned several orders of magnitude; therefore normalized fluorescence is shown on a 2 7 logarithmic scale. The Ca abundance of the epidermal cell layer of the embryo was lower 2 8 than the internal cell layers. This distribution was also observed for P (See Supplemental 2 9 Figure 2). In wild type embryos, Ca abundance in cells of the endodermis (marked on 3 0 Figure 3B) was comparatively greater than neighboring layers; the pericycle, protoxylem and protophloem on the interior and cortical cells on the exterior ( Figure 3B). The 1 endodermal cells contain storage vacuoles rich in Fe (Roschzttardtz et al., 2009), which 2 corresponded to enrichment of Fe, Ca, Mn and Zn. This tissue-level Ca gradient was less 3 pronounced in 35s-sCAX1 ( Figure 3D and E) and absent in cax1cax3 ( Figure 3H). Given 4 the Ca enrichment of the endodermis, and the likelihood that endodermal organelles are 5 storage vacuoles, we chose to further analyze this layer for CAX phenotypes.

6
Looking closely at individual endodermal cells ( Figure 3C, F and I), Ca appeared 7 to be highly localized within the lumen of a subcellular organelle in wild type. This is 8 indicated by the increased Ca abundance at the center of the body in comparison with a 9 lower abundance at the margin. This contrasted with lines expressing an activated CAX1 1 0 (35s-sCAX1) where Ca was localized at higher abundances at the margin (arrows, Figure   1 1 3F). In this line, Ca appeared to be associated with numerous bodies of generally smaller 1 2 size ( Figure 3F) as well as with irregular shaped masses that could indicate its presence in Col-0. Figure 3I shows both an endodermal (en) and cortical cell (co), and organelles indicates that these elements were not confined within organelles in cax1cax3 2 3 endodermal cells in the same manner as wild type cells.

4
High resolution images were conducted at an energy (10 keV) and a resolution 2 5 that made it possible to see other elements of biological interest within the cell, such as P, 2 6 S, Fe and Cu. Given the novelty of subcellular resolution images of the seed, we 2 7 examined the characteristic elemental distributions in these sections. We found common 2 8 distributions of elements; namely that certain elements were either located within the 2 9 lumen of subcellular organelles (e.g. in wild type P, Fe and Ca), outside of organelles 3 0 (e.g. S) or associated with the cell wall or cell membrane (e.g. Cu and Ca) (Figure 4).
High resolution images were also collected of whole-seed sections to gain an 1 understanding of elemental association with certain tissue types. Recent studies suggest 2 that only certain cells have the ability to accumulate Ca (Conn et al., 2011), but this has 3 not previously been imaged at this resolution in Arabidopsis seed for a range of Mn bulk and spatially resolved concentrations, we wanted to look at Mn in greater detail.

0
We found in our earlier studies that Mn was strongly localized to a sub-epidermal layer  two synchrotron x-ray fluorescence microprobes to collect high resolution spatially 2 5 resolved elemental images of how CAX transporters impact nutrient distribution within 2 6 seeds.

7
For technical reasons, we focused our SXRF analysis on seeds. Seed is naturally 2 8 dehydrated and has an extended stability during analysis and is therefore ideally suited to 2 9 x-ray imaging studies. Numerous experiments have used transgenic approaches to alter rice nutrient content (Lee et al., 2009) and SXRF can now be used to analyze the spatial 1 distribution of nutrients within seeds.
2 Some of ours observations in seeds may not be applicable to other parts of the 3 plant (Vreugdenhil et al., 2004). For example, in seeds some nutrients may be in a 4 complex with the P containing compound phytate. Elemental distribution patterns may 5 vary among tissues and it is likely that the correlations observed here would not be found 6 in other tissues. however, this does not appear to drastically alter the P levels in the seeds.

2
The distributional changes observed (increases in Ca outside of organelles) are  coat is derived from maternal tissues (the integuments of the ovule), and in this study 1 7 differences in Ca concentration of the embryo and the seed coat were found in CAX 1 8 mutants.

9
The semi quantitative RT-PCR provides a snapshot of the transcript levels of 2 0 CAX1 and CAX3 in mature dry seed. At this stage, the level of CAX1 transcript is low, transcripts at earlier developmental periods may influence the transport activity that  there is the potential to use elemental imaging to explore gene × environment interactions 2 4 by imaging samples with altered gene expression under various environmental stresses.

5
With advances in detector technology, the collection of in vivo data will remove the need 2 6 for intrusive sample preparation in the majority of instances. However, even with faster 2 7 detectors it is likely that radiation damage from x-rays will limit the possibility of in vivo it is necessary also to increase the bioavailable form of the nutrient rather than simply 7 increasing the bulk amount. Our long-term goal is to combine ionomic and imaging 8 approaches to identify the relationships between nutrient distribution and subsequent 9 changes in the chemical forms of nutrients in the plant cell.  cax1-1, cax3-1, cax1cax3 35S-sCAX1 and wild type plants were grown as 2 7 described above, in the same growth chamber and harvested on the same day. Three replicates of approximately 50 mg aliquots of dry seed were digested in 2 ml Optima 2 9 HNO 3 in Teflon vessels using a MARS5 EXPRESS microwave-assisted reaction system 3 0 (CEM, Mathews, NC). A standard reference material (NIST 1573a, Tomato Leaves) and a HNO 3 blank were included after every fifth sample. Sample volume was brought to 10 1 ml with DI water. The vessels were heated to 180°C in 10 minutes and held at that 2 temperature for a further 10 minutes. After the samples had cooled, they were brought up 3 to approximately 15 ml volume with DI water. Samples were analyzed for trace element 4 concentrations using an Agilent 7500cx ICP-MS operating in collision mode at the Trace 5 Element Analysis Core facility of Dartmouth College. Tomograms were collected at the bending magnet beamline X26A at the National 2 Synchrotron Light Source, Brookhaven National Laboratory (Upton, NY). X-ray 3 fluorescence measurements were conducted using a 12 keV monochromatic x-ray beam.

4
Due to spectral overlap with the abundant macronutrient K, fluorescence data could not 5 be collected for Cd at the 12 keV excitation energy. Monochromatic x-rays were tuned 6 using a Si (111) channel-cut monochromator, and focused to a beam size of 5 × 8 µm 7 using Rh-coated, silicon Kirkpatrick-Baez microfocusing mirrors. Incident beam energy 8 was monitored using an ion chamber upstream of the focusing optics. X-ray fluorescence 9 spectra were collected with a Vortex-EX silicon-drift detector (SII Nanotechnology) with 1 0 an active area of 50 mm 2 . X-ray transmission through the sample was recorded 1 1 simultaneously using a p-type, intrinsic, n-type (PIN) photodiode.

2
Individual mature, dry and unsectioned seeds were attached to a 100 µm diameter 1 3 silica fiber using Devcon® 5-minute epoxy resin, with the micropyle uppermost. The 1 4 fiber was inserted in to a Huber 1001 goniometer, mounted on a xyzθ stage and centered. detector, required that seed be sectioned for analysis, which necessitated a resin-2 2 embedding sample preparation step. Arabidopsis seed was imbibed on moist filter paper 2 3 for two days to allow removal of the seed coat and release the embryo. This was carried 2 4 out to ensure optimal infiltration of resin in to the embryo cells. Embryos were placed in minutes each) with the final step repeated three times over the course of one hour.

0
Embryo samples were then immersed in three changes of 100% ethanol for ten minutes each, followed by a LR White resin: ethanol mixtures of 1:3, 1:2 and 1:1 (twice) for one 1 hour each, after which they were stored at 4˚C in 1:1 LR White resin: ethanol overnight.

2
Samples were warmed to room temperature and moved to a 2:1 LRWhite:ethanol 3 solution over four hours, before immersing in two changes of 100% LR White solution 4 for one hour each. Samples were stored at 4˚C overnight, and then warmed to room 5 temperature the following day before immersing in three changes of 100% LR White 6 resin over a 4 hour period. Embryos were transferred to flat embedding moulds using a 7 toothpick to achieve the correct orientation, before polymerizing for 24 hours. Cleveland, OH) because both satisfied our size requirements, were easy to work with, 1 3 were low cost and withstood exposure to LR White. These polycarbonate film was cut 1 4 using a scalpel to fit standard 25 3 75 mm 2 glass microscope slides. The slides were pre- one side of the gasket and pressed on to the slide so an airtight seal was formed. Silicone 2 0 adhesive was cured at room temperature for 24h or in a 60˚C oven for one hour and 2 1 allowed to cool before use.

2
LR White and embryo specimens were added until the chamber was slightly over 2 3 filled and the liquid formed at convex surface, and then an Aclar strip (Ted Pella) was cut 2 4 to a slightly larger width than the slide and placed on top of the chamber to shield the LR 2 5 White from oxygen. One end of the Aclar strip was placed on to the edge of the chamber 2 6 and the rest the rest rolled down on to the resin so that any excess resin spilled over the 2 7 side and prevented air bubbled from being trapped underneath. The specimens were 2 8 polymerized at 60˚C for 24 hours. After polymerization, the Aclar was removed and 2 9 specimens were excised with a razor blade while the slide was still warm. Initial samples were cut to between 1-5 µm thick with a microtome and a glass knife and allowed to 1 adhere to silicone nitride windows  for each element are scaled to the highest maximum pixel abundance, shown on the 1 7 colorbar, expressed as moles, the minim is zero for all elements.