Plant growth
Alpine Penny-cress (Noccaea caerulescens (J.Presl & C.Presl) F.K.Mey = formerly Thlaspi caerulescens J.Presl & C.Presl), ‘Ganges’ ecotype, was germinated as described earlier [25] and then grown hydroponically for 12 weeks before starting to harvest the leaves for the experiments. The nutrient solution HHNS (“hyperaccumulator hydroponic nutrient solution”) consisted of 1000 μmol L−1 Ca(NO3)2, 500 μmol L−1 MgSO4, 50 μmol L−1 K2HPO4, 100 μmol L−1 KCl, 10 μmol L−1 H3BO3, 0.1 μmol L−1 MnSO4, 0.2 μmol L−1 Na2MoO4, 0.1 μmol L−1 CuSO4, 0.5 μmol L−1 NiSO4, 20 μmol L−1 Fe(III)-EDDHA (Fe(III)-ethylenediamine-di(o-hydroxyphenylacetic acid), and 100 μmol L−1 ZnSO4 as described before [25]. Plants were grown in a greenhouse with automatic ambient light supplementation by a 1:1 mix of cool white and warm white LEDs (Photon System Instruments, Brno, Czech Republic) to achieve a 16 h sinusoidal light cycle with 500 µmol m−2 s−1 photon flux density. The temperature was regulated to max. 25 °C during the day. Leaves from three plants were sampled from five different points of the rosette covering different stages of development: from the young leaves in the apical meristem rosette down to the elderly, but not senescent, leaves. The harvesting strategy along the rosette is shown in Fig. 1.
Soybean (Glycine max (L.) Merr, cultivar ‘Galina’) seeds were germinated on a moistened mixture of perlite and vermiculite (1:3) in the dark at 25 °C to promote radicle development (similarly to e.g. Graham [26]). After 3 days they were transferred to the phytochamber with a sinusoidal light cycle (maximum 500 µmol m−2 s−1, provided by 8 channels of LEDs to simulate sunlight; Photon System Instruments, Brno, Czech Republic), temperature cycle (25 °C at noon and 18 °C at night) and moisture cycle (40% relative humidity in the afternoon, 60% in the morning before onset of light). Ten days old seedlings (four plants per pot) were transferred to modified ½ strength of the HHNS [10, 25] consisting of 1000 μmol L−1 Ca(NO3)2, 250 μmol L−1 MgSO4, 250 μmol L−1 K2HPO4,50 μmol L−1 KCl, 10 μmol L−1 H3BO3, 0.5 μmol L−1 MnSO4, 0.2 μmol L−1 Na2MoO4, 0.3 μmol L−1 CuSO4, 1 μmol L−1 ZnSO4, 0.5 μmol L−1 NiSO4 and 20 μmol L−1 Fe(III)-EDDHA (Fe(III)-ethylenediamine-di(o-hydroxyphenylacetic acid). The nutrient solution was continuously renewed with a flow rate of 150 mL day−1 plant−1, which was increased to 250 mL day−1 plant−1 after 1 week of treatment as described before [25]. After 3 weeks, the second root branch from the top was taken for analysis, and after rinsing in Zn-deficient HHNS the middle part of the root branch was placed in the measuring chamber filled with Zn-HHNS.
Haller’s rockcress (Arabidopsis halleri (L.) O’Kane & Al-Shehbaz) plants were grown on soil (mix of 70% commercial peat-free gardening soil and 30% sand) with the same light and temperature conditions as described for N. caerulescens.
Chili pepper (Capsicum annuum, cultivar ‘Kozí Roh’) plants were grown on soil (mix of 70% commercial peat-free gardening soil and 30% sand) under natural daylight and room temperature. Young-mature pepper leaves (just having reached the final size of 3–4 cm) were harvested from approximately 3 to 4 years-old plants.
Determination of total metal concentration with ICP-MS
For total metal concentration the lyophilized N. caerulescens and C. annuum leaves (about 30 mg) were digested in glass tubes with 0.5 mL of a mixture of 85 L 100 L−1 of concentrated (70%) HClO4 (Suprapur® grade, Carl Roth, Karlsruhe, Germany and 15 L 100 L−1 of concentrated (69%) HNO3 (Ultrapur® grade, Carl Roth, Karlsruhe, Germany) following the protocol of Zhao et al. [27]. Glycine max roots (four plants per pot) were washed twice with double distilled water (ddH2O), tap dried and homogenized in liquid nitrogen. Following lyophilization, about 30 mg of the pooled root samples were used for digestion in three technical replicates.
The glass tubes were uniformly heated using a Fuji PXG4 Thermoblock (AHF Analysentechnik AG, Tübingen, Germany). A program was used for ramping the temperature of the acid mixture to 220 °C for 4 h. The acid mixture was then heated at 220 °C for another hour to dry the digest from acid contents. The digest was then cooled to room temperature and 0.5 mL of 5% HCl (Ultrapur® grade, Carl Roth, Karlsruhe, Germany) was added to each test tube. Afterwards the glass tubes were heated to 90 °C for 1 h to obtain clear solutions. The final volume was made to 1.5 mL with ddH2O and stored in 2 mL microcentrifuge vials. Sample solutions were diluted 7000× with 0.2% HNO3. Indium was added as internal standard at 1 µg/L to each test solution. The Inductively Coupled Plasma (ICP) multi-element standard solution VI (Merck KGa, Darmstadt, Germany) was used for preparation of several ranges of calibration points. The solution was chosen as standard since the matrix effect was negligible for the quantification of analytes in plant digests. The sector field Inductively Coupled Plasma Mass Spectrometry (ICP-sfMS) Element XR-2 with jet interface (Thermo Fisher Scientific, Bremen, Germany) was used for the elemental analysis of the plant digests. The instrument was optimally tuned to reduce the potential interferences by choosing low, medium and high resolutions (similar to Andresen et al., 2013 [28]). Oxide formation rate was acceptably low as monitored by CeO+/Ce+. The typical operating conditions of the ICP-sfMS were: RF power: 1250 W, spray chamber temperature: 2 °C, oxide ratio CeO+/Ce+: 1.0–1.2%, Doubly charged Ce2+/Ce+: 1.0–1.2%, auxiliary Gas: 0.8 L min−1, Sample gas flow: 1.20 L min−1 (variable), Cool gas: 16 L min−1, Extraction lenses: − 2000 V, Low resolution: 300, Medium resolution: 4000, High resolution: 10,000, Interface cones: Ni sample and H-skimmer cones. In order to avoid contaminations, the ICP-MS is located in a cleanroom with air filtration. All measurements follow a “metal free” protocol, proved in earlier works involving metal deficiency [29], which forbids metal tools and is using Polytetrafluoroethylene (PTFE) tools and vials made from PFA (Perfluoroalkoxy alkanes; Savillex, Eden Prairie, MN, USA) because of their low metal contamination.
Mounting of intact samples (leaves and roots) for in vivo measurements
For all in vivo measurements (chlorophyll fluorescence kinetics and µXRF), samples were mounted in a modified version of the measuring chamber that was originally developed for fluorescence kinetic measurements [25]. The new version of the chamber is shown in Fig. 2. Compared to the previously published version of the chamber, the distance from the sample to the highest point of the chamber had to be drastically reduced because of the short working distance of the benchtop system; when in focus it leaves about 6 mm space between the crash protection plate and the surface of the sample. Therefore, the inlets and outlets for air/nutrient media of the chamber had to be moved from the top to the side, the lid had to be made thinner, and the toric seal (O-ring) holding the leaf assembly had to be moved directly to the lower side of the lid. Finally, the glass window was replaced by a window made of 80 µm thick laser printer foil to diminish the absorption of X-rays.
Leaves were mounted by putting them onto the measuring chamber, then placing a cotton pad on top, then a plastic disc (with holes for aeration) for gently pressing them. The whole assembly was finally covered with a fine (5 µm pore size) nylon mesh from below, which was stretched by an O-ring (Fig. 2 and Additional file 1: Figure S1). The petiole of the leaf was put into a small water-filled Petri dish in the bottom of the chamber, to keep the leaf hydrated. The chamber was flushed with air. The air was saturated with water by making it bubble in a flask filled with water with an aquaria pump (TetraTec APS 300, Tetra, a Spectrum Brands Company, Melle, Germany) and connected with PTFE tubing to the inside of the µXRF machine enclosure and to the measuring chamber inside (Fig. 3).
Root sections were mounted by putting them in nutrient solution without Zn onto the window of the chamber. Afterwards, a piece of cellophane previously boiled and washed several times in ddH20 to remove any adhering particles was stretched over the roots by the O-ring. The chamber was filled with the same nutrient solution, which was continuously renewed during the measurement via a peristaltic pump (Ismatec REGLO ICC Digital Peristaltic Pump, Cole-Parmer GmbH, Wertheim, Germany) attached to the PTFE tubing outside of the hutch of the µXRF machine (see Fig. 3).
Chlorophyll fluorescence kinetic measurements
Chlorophyll fluorescence kinetics (for more information see Genty et al. [30]; Baker [31]; Maxwell and Johnson [32]; Stirbet and Govindjee [33]) was measured using a macroscopic fluorescence imaging system with a newly developed ultrafast camera and software for direct imaging of fast fluorescence transients (Photon Systems Instruments (PSI), Brno, Czech Republic). All technical properties of the instrument are described in a recent publication [34]. Measurements of fast chlorophyll fluorescence transient (OJIP) induction kinetics were conducted using a custom-made protocol on dark-adapted leaves as described by Küpper et al. [34]. The frame period was 250 µs, the shutter opening (measuring flash length) was 100 µs and the supersaturating pulses were 4000 µmol m−2 s−1. The definitions of OJIP parameters as described by [33] were used for calculating maps of OJIP parameters: ΦPo—maximum quantum yield of primary PSII photochemistry, ΦET2o—quantum yield of the electron transport flux from QA to QB, and ΦRE1o—quantum yield of the electron transport flux until PSI electron acceptors. Quantum yield refers to stable charge separation or oxygen evolution divided by the number of absorbed photons, or the efficiency of photochemistry (Kalaji et al. [35] and references therein).
Measurements of slow chlorophyll fluorescence induction and quenching kinetics were done according to the protocol described by Küpper et al. [25] but with adaptations for the new measuring device as described by Küpper et al. [34]. In summary, a 1000 ms flash of supersaturating light (4000 µmol m−2 s−1) for Fm was followed by 90 s of darkness, after which F0 was measured for 5 s. Then, 100 s of actinic light were applied to analyse the Kautsky induction, and finally 100 s of measurement with no actinic light were used to measure dark relaxation and \( {\text{F}}_{0}^{{\prime }} \). During the actinic light exposure (100 µmol m−2 s−1) and in the dark relaxation period, 600 ms supersaturating flashes were applied for analysis of photochemical (ΦPSII-operating efficiency of PSII) and non-photochemical quenching (NPQ = (Fm − \( {\text{F}}_{\text{m}}^{{\prime }} \))/Fm). A typical Chl fluorescence trace is shown in Additional file 2: Figure S2.
Chlorophyll fluorescence kinetics of pepper leaves was measured immediately before and immediately after the µXRF measurements, which lasted about 20 h. OJIP (ΦPo, ΦET2o and ΦRE1o) and Kautsky parameters (ΦPSII and NPQ) were compared between the X-ray exposed area and adjacent, bordering part of the leaf of the same size as the exposed area. During the µXRF measurements, the camera light was switched off, but the measuring hutch was illuminated.
X-ray microfluorescence scanning
X-ray fluorescence was measured with a customised M4 TORNADO system (Bruker Nano GmbH, Berlin, Germany). This machine was equipped with two XFlash® silicon drift detectors (type SDD VH50P) able to cope with count rates as high as 310,000 counts per second (cps) with an energy resolution < 145 eV (Mn Kα) according to manufacturer’s parameters (Bruker Nano GmbH, Berlin, Germany). In this work all measurements were done with the Rh tube with a 0.1 mm Be window (Incoatec GmbH, Geesthacht, Germany) and polycapillary optics (IFG Institute for Scientific Instruments, Berlin, Germany). It was operated at 50 kV and 600 µA. The polychromatic incoming flux of the X-ray tube spans over 40 keV with the structure shown in the Additional file 3: Figure S3. All photons with energy above the chosen absorption edge will not be equally absorbed, but with an energy dependent cross section [36]. This means that the beam is not equivalent to a monochromatic beam with the same flux (photons/second). The measuring hutch and the sample stage with the in vivo measuring chamber mounted on it are shown in Fig. 3. A filter composed of 100 µm of Al and 25 µm of Ti was used to flatten the bremsstrahlung in the energy region of the transition metals. A polycapillary lens was used to focus the photons with the highest transmission at 9 keV [37]. Besides the measurement of the transmission of a polycapillary lens shows little variation in the range 5–10 eV (see Fig. 1 in Wolff et al. [38]). The polycapillary lens provided a beam spot size of about 15 µm for the Mn Kα line. The machine was customized for the needs of this project in several ways: (1) At request of the authors the machine was specially fitted with a multilayer mask (Bruker Nano GmbH, Berlin, Germany) on the detectors to reduce spurious counts of Ni, Cu and Zn, resulting in a clean spectrum of Zn and Cu, with a few % contamination of Ni (Additional file 4: Figure S4) [39]. (2) In order to be able to measure samples in vivo, a PTFE tubing was inserted that transports water-saturated air or nutrient solution through the plant measuring chamber. (3) Since the measuring chamber cannot be operated in vacuum, a line for He flushing (inlet and outlet) was installed for enhancing the measurement of low-Z elements. The fluorescence maps were collected for a total time of 3–25 ms per pixel for N. caerulescens, 150 ms for A. halleri and 480–720 ms for pepper and soybean.
Resolution of the X-ray microfluorescence scanning
When measuring elements from the soft to the hard X-ray range, it is important to remind that the spot size is energy dependent [37] with a term that is inversely proportional to the energy [24]. Smaller spot sizes are obtained for increasing energies. The size of the spot for transition metals was about 15 µm. To know in how far the spot size is influenced by primary emission filters, the edge of an aluminium foil was scanned without and with an Al–Ti filter (Additional file 5: Figure S5). This test, which also included the absorption and reflectivity effects, showed that the effect on the resolution of the polychromatic X-ray source was negligible.
Standards for quantification
The foil standards for calibration were made to include all elements of interest for the current study, with the concentration ratios approximately (order of magnitude) matching the plant samples. Foil standards matching the thickness of the plant leaves were made by dissolving polyvinyl alcohol (PVA, Ultimaker, Geldermalsen, The Netherlands) in double distilled water (20% w/v). PVA was chosen as a matrix because among available polymers it (a) resembles an organic matrix quite well with its formula [CH2CH(OH)]n, (b) it is very well water soluble and therefore homogenously miscible with the metal solutions until dryness. To this solution, Ca(NO3)2 and KCl were added from 100 mmol L−1 stock solutions, ZnSO4 was added from a 1 mol L−1 stock solution, Cu-EDTA, FeNa-EDTA, Mn-EDTA and Ni-EDTA were added from 10 mmol L−1 stock solutions. The residual volume was filled with water. The resulting final concentrations in the standard stock were: 10% PVA, 0.5 mmol L−1 Cu and Mn, 1 mmol L−1 Fe, Na and Ni, 5 mmol L−1 Ca and K, 5.5 mmol L−1 Cl, 11.5 mmol L−1 S, 100 mmol L−1 Zn (for N. caerulescens; otherwise 0.5 mmol L−1 Zn). The mixture was diluted with 10% PVA to yield standards of different concentrations. The standards were pipetted into a 3D-printed grid of 1 × 1 cm2 squares, and left to dry in a cleanroom. Foil layers were stuck to reproduce 200 and 400 μm thickness. The metal content of the foil was determined with ICP-MS. Standards for root samples were polyimide tubings of a diameter close to the root of interest, filled with aqueous solutions of all elements of interest plus 20% glycerol to bring the content of light elements (C, H, O) close to a plant sample.
Quantification of a N. caerulescens leaf µXRF maps with empirical standards
The fluorescence maps were processed using the software provided with the machine. The fitting method was created with the Xmethod software [40] (Bruker Nano GmbH, Berlin, Germany). The background is determined mathematically by a peak stripping approach that gives a flat background after several cycles. The data was calibrated with the mentioned foil standards to a straight line without offsets. This fully empirical method relies on the assumption that the foil standard resembles the sample matrix as well as the sample thickness. Two different sets of standards were used to match the thickness of about 200 µm for young leaves (at position 0) and 400 µm for young mature to fully mature leaves (1–4) of N. caerulescens, with thicknesses measured with a digital caliper (model Workzone, Dario Markenartikelvertrieb GmbH and Co KG, Hamburg, Germany) and compared to earlier work [18]. The element maps of an entire leaf were collected without sample damage after 4–20 h of measurement with a maximum dwell time of 600 ms in each spot. For full quantification the N. caerulescens pixels were binned to the lowest amount the software could process (5 × 5 to 9 × 9). The reduced resolution did not affect the conclusions since the epidermal cells in N. caerulescens have sizes in the range of 100 µm [18, 41]. The µXRF maps of crop plants (C. annum and G. max) were not fully quantified, but shown as semi-quantitative maps after spectral deconvolution, in order to avoid the pixel binning and corresponding loss of resolution. Their measured areas were smaller (5 × 5 mm for leaves, and up to 3 × 2 mm for roots) so that the number of pixels was not a problem, but Gaussian smoothing was needed to reduce noise. Gaussian smoothing and assignment of colour scales were done in the Fiji version of ImageJ [42]. Maps were assembled, arranged and labelled using PhotoImpact X3 (Corel Corporation, Ottawa, Canada).
Samples were measured in triplicates for the five developmental stages of N. caerulescens leaves. Positions 2 and 3 were later averaged due to their similarity. Mean values and standard deviations were calculated in Numbers (macOS High Sierra 10.13.6, Apple, Cupertino, CA, USA). Four leaves from four independently grown plants were measured from C. annuum. The root of G. max was measured 2×, once the emerging root hair and once the tip of the lateral root.
Statistical analysis
Statistics was analysed in Origin Professional (versions 2015 and 2019; Originlab Corporation, Northampton, MA, USA). For analysis of chlorophyll fluorescence parameters that are saturating and do not have normal distribution (ΦPo, ΦRE1o, ΦEt2o and ΦPSII), the non-parametric Mann–Whitney U test [43] was used for comparison of the X-ray exposed and the adjacent leaf area (same size, bordering the exposed area). One-way repeated measures ANOVA and pair-wise comparison (Bonferroni test) were used to distinguish the effects of X-ray on the same leaf measuring area by comparing the values of NPQ_i1 before and after the X-ray exposure. This test deals with dependent variable which is subjected to repeated measurements—the independence assumption is not considered. The significance level of 0.05 was applied.