Plant material and growth conditions
In vitro micropropagated [26] Populus × canescens (INRA717 1-B4) plantlets were grown on half-strength Murashige and Skoog medium for three weeks to develop roots. Afterwards, they were planted singly into PVC tubes (5 cm diameter; 40 cm length) with one drain [27, 28] filled with autoclaved sand (Ø 0.4–3.15 mm particle size, Melo, Göttingen, Germany) and grown in a greenhouse at an air humidity of about 65 %. The plants were supplemented with additional light (EYE Clean Ace MT400DL/BH, EYE Lighting Europe, Uxbridge, UK; 50–100 μmol quanta m−2 s−1of photosynthetically active radiation, depending on poplar height and natural light condition) for 14 h a day from 7 am to 9 pm. The plants were automatically irrigated as described by Müller et al. [28] every 4 hours (ca. 6.5 mL, after 45 days ca. 9 mL) with Long Ashton nutrient solution [29] containing either high phosphate (HP) supply (200 μM KNO3, 900 μM Ca(NO3)2, 300.2 μM MgSO4, 599.9 μM KH2PO4, 41.3 μM K2HPO4, 10 μM H3BO3, 2 μM MnSO4, 7 μM Na2MoO4, 40 nM CoSO4, 200 nM ZnSO4, 128 nM CuSO4, 10 μM EDTA-Fe, in total: 641 μM Pi) or reduced Pi concentrations. Mildly Pi starved (medium phosphate, MP) poplars received Long Ashton solution with 5.999 μM KH2PO4 and 0.413 μM K2HPO4 and additionally 675.8 μM KCl and Pi starved (low phosphate, LP) plants 0.060 μM KH2PO4 and 0.004 μM K2HPO4 and additionally 682.5 μM KCl. Plant height was measured weekly from the stem base to the apex. This experiment was repeated four times for different measurements with 12 biological replicates each time.
Labeling of the poplars with 33P and harvest
Sixty-day-old HP, MP and LP poplars (n = 5 per treatment, experiment 2) were watered by hand at 7 am and 10 am with the respective nutrient solution (16 and 14 mL for HP, 4 mL for MP and 2 mL for LP) and labeled with H3
33PO4 (Hartmann Analytic, Braunschweig, Germany) at 11 am: 20.25 μL of the H3
33PO4 stock solution were mixed with 5.5 mL of each of the respective nutrient solutions and 1 mL of each labeling solution was applied once to the poplars. This treatment resulted in 1.2 (±0.013 resp. 0.002) MBq for the HP and MP plants and 1.12 (±0.034) MBq for the LP plants (mean ± SE, n = 3). The specific radioactivity was 1.88 × 103 Bq nmol−1 P for HP, 1.87 × 105 Bq nmol−1 P for MP and 1.46 × 107 Bq nmol−1 P for LP poplars.
The automatic irrigation was stopped during the chase period of two days. During this time the poplars were irrigated by hand with unlabeled nutrient solution avoiding through-flow. Two days after label application, the plants were harvested. The roots were briefly washed with tap water. Each plant was divided into fine roots, coarse roots, stem and leaves. The biomass of the tissues was determined immediately after harvest and after drying at 60 °C for 7 days.
Phosphorus distribution at the whole-plant level
To visualize the distribution of radioactivity, the poplars were dried at 60 °C for one day pressed between paper and two glass plates. Autoradiographs were taken with a Phosphorimager (FLA 5100, Fuji, Japan) after exposure for 30 min on imaging plates (Imaging Plate BAS-MS 2040, 20 × 40 cm, Fuji, Japan). The image was taken with the program Image Reader FLA-5000 (version 3.0, Fuji Film, Japan) with 100 μm resolution and analyzed with AIDA Image Analyzer (version 4.27, raytest Isotopenmeßgeräte, Straubenhardt, Germany).
Determination of net 33P and total P uptake
Dried plant tissues were milled (Retsch, type MM2, Haan, Germany) to a fine powder. About 25 mg of leaf or stem powder and 10 mg of fine or coarse root powder were weighed into glass vials (20 mL, PerkinElmer Life and Analytical Sciences, Rodgau, Germany) and incinerated for 4 h at 500 °C (M104, Heraeus Holding, Hanau, Germany). The residual material was mixed with 10 mL of scintillation liquid (Rotiszint® eco plus [Carl Roth, Karlsruhe, Germany]) and the radioactivity was measured by liquid-scintillation technique (Tri-Carb® 2800TR [PerkinElmer Life and Analytical Sciences, Rodgau, Germany]). The maximum count time was 10 min in normal count mode calculated against a quench set of 33P. All values were corrected for the half-life of 33P. 33P-activity in plants and proportion of 33P-uptake were calculated using following equations:
$$ \begin{array}{l}\mathrm{Radioactivity}\ \mathrm{in}\ \mathrm{plant}\ \mathrm{tissue}\ \left[\mathrm{Bq}\right]=\\ {}\left(\mathrm{activity}\ \mathrm{in}\ \mathrm{vial}\ \left[\mathrm{Bq}\right]\hbox{-} \mathrm{background}\ \left[\mathrm{Bq}\right]\right)\times 0{.5}^{\frac{\mathrm{date}\ \mathrm{of}\ \mathrm{harvest}\hbox{-} \mathrm{date}\ \mathrm{of}\ \mathrm{measurement}}{25.3}}\times \frac{\mathrm{dry}\ \mathrm{mass}\ \mathrm{tissue}\ \left[\mathrm{g}\right]}{\mathrm{sample}\ \mathrm{in}\ \mathrm{vial}\ \left[\mathrm{g}\right]}\end{array} $$
$$ {}^{33}\mathrm{P}\hbox{-} \mathrm{recovery}\ \left[\%\right]=\frac{\mathrm{radioactivity}\ \mathrm{in}\ \mathrm{plant}\ \left[\mathrm{Bq}\right]}{\mathrm{given}\ \mathrm{radioactivity}\ \left[\mathrm{Bq}\right]}\times 100 $$
To determine total P uptake the specific radioactivity of the nutrient solution was used:
$$ \mathrm{P}\ \mathrm{uptake}\;\left[\mathrm{nmol}\right]=\frac{\mathrm{measured}\ \mathrm{radioactivity}\ \mathrm{in}\ \mathrm{plant}\ \left[\mathrm{Bq}\right]}{\mathrm{specific}\;\mathrm{radioactivity}\;\left[\mathrm{Bq}\;\mathrm{n}\mathrm{m}\mathrm{o}{\mathrm{l}}^{\hbox{-} 1}\mathrm{P}\right]}. $$
Determination of total P contents
Dry fine root, coarse root, leaf and stem of 63-day-old HP, MP and LP poplars (n = 4 per treatment, experiment 1) were powdered and pressure-extracted in HNO3 [30]. Total phosphorus concentration was measured using an inductively coupled plasma optical emission spectrometer (ICP-OES; Optima 5300 DV, PerkinElmer Life and Analytical Sciences, Rodgau, Germany). P use efficiency (PUE) was calculated as dry mass per total P content. The equation for the whole plant PUE was
$$ \mathrm{P}\mathrm{U}\mathrm{E}\ \left[{\mathrm{g}\ \mathrm{mg}}^{\hbox{-} 1}\right] = \frac{{\mathrm{DM}}_{\mathrm{total}}}{\mathrm{P}\ {\mathrm{conc}}_{\mathrm{L}}\times {\mathrm{DM}}_{\mathrm{L}}+\mathrm{P}\ {\mathrm{conc}}_{\mathrm{S}}\times {\mathrm{DM}}_{\mathrm{S}}+\mathrm{P}\ {\mathrm{conc}}_{\mathrm{FR}}\times {\mathrm{DM}}_{\mathrm{FR}}+\mathrm{P}\ {\mathrm{conc}}_{\mathrm{CR}}\times {\mathrm{DM}}_{\mathrm{CR}}} $$
with DM: dry mass [g], P conc: total P concentration in fraction [mg g−1], L: leaves, S: stem, FR: fine roots, CR: coarse roots.
Phosphate uptake after glucose supply and determination of carbohydrates
Seventy-seven-day-old HP, MP and LP poplars (n = 5 per treatment, experiment 3) were irrigated with 80 mL HP, MP or LP Long Ashton solution with or without 400 mM glucose at 5 am. Along with turning on the additional light at 7 am, they were labeled with H3
33PO4 (Hartmann Analytic, Braunschweig, Germany) in 1 mL of HP, MP or LP Long Ashton solution to yielding 1.08 (±0.006) MBq in the growth medium of HP poplars and 1.09 MBq (±0.0008 resp. 0.003) in the growth medium of MP and LP poplars (mean ± SE, n = 2). At the end of the light period (9 pm), the poplars were harvested to determine biomass, 33P activity (as above), and the carbohydrate concentrations. Leaves (half of each leaf) and fine roots for carbohydrate determination were immediately shock frozen in liquid nitrogen and stored at −80 °C. The tissues were freeze-dried at −80 °C for three days (BETA1, Martin Christ Gefriertrocknungsanlagen, Osterode am Harz, Germany) and milled (as above). About 25 mg tissue powder was extracted in 1.5 ml dimethyl sulfoxide/hydrochloric acid (80:20 (v:v)) at 60 °C for 30 min. After centrifugation, 200 μL of the supernatant were mixed with 1250 μL 0.2 M citrate buffer (pH 10.6). After an additional centrifugation, 400 μL of the supernatant were mixed with 400 μL citrate buffer (50 μM, pH 4.6). Two hundred μL were mixed with 250 μL reaction solution (4 mM NADP, 10 mM ATP, 9 mM MgSO4, 0.75 M triethanolamine, pH 7.6) and 300 μL H2O. The reaction was started by subsequent addition of enzymes and the production of NADPH was determined photometrically at a wavelength of 340 nm and 25 °C until no further increase was observed. For glucose determination 10 μL (ca. 3.4 U; 1.7 U) of hexokinase; glucose-6-phosphate-dehydrogenase (Roche Diognostics, Mannheim, Germany), and subsequently for fructose 5 μL (ca. 17.5 U) of phosphoglucose-isomerase (Roche Diognostics, Mannheim, Germany) were added. To determine the amount of sucrose 400 μL (ca. 12 U) invertase (Sigma-Aldrich Chemie, Steinheim, Germany) solution (0.1 mg invertase in 1 mL 0.32 M citrate buffer, pH 4.6) were mixed with 400 μL of supernatant and then glucose was determined as above. Free glucose was subtracted. The concentration of glucose (fructose) was calculated for ΔE = E1-E0 with E1 and E0 being the extinction of assay after and before addition of glucose-6-phosphate-dehydrogenase (after and before addition of phosphoglucose-isomerase) and the extinction coefficient ε = 6.3 L mmol−1 cm−1 for NADPH. Sucrose concentration was below the detection limit in most of the samples. Each sample was measured in triplicate.
Kinetic measurements
Roots of nine-week-old HP, MP and LP poplars (experiment 4) were washed carefully with tap water to remove the sand. The plants were transferred to aerated nutrient solution with HP, MP and LP Long Ashton nutrient solution and acclimated to this and lab conditions under continuous light (about 10 μmol quanta m−2 s−1 of photosynthetically active radiation, TLD18W/840, Philips Lighting, Hamburg, Germany) for 1 day. Phosphate uptake was determined with the method of Claassen and Barber [31]. The plants were removed from the nutrient solutions, the roots were cautiously surface-dried between tissue papers, washed with the experimental solution (see below), surface-dried again and placed in plastic beakers. An appropriate volume (5 to 45 ml) of Long Ashton nutrient solution which contained 213.7 μM Pi for all plants and 33P-phosphoric acid (Hartmann Analytic, Braunschweig, Germany) (to about 5375 Bq mL−1 of experimental solution) was added. During the time course of the experiment of up to 14 h, the nutrient solution was stirred and aerated with compressed air and water loss by plant transpiration was replaced by adding deionized water up to the marked original level. Uptake of P was calculated by the decrease in 33P in the experimental solution. For this purpose, 40 μL of experimental nutrient solution was removed at distinct time intervals and mixed with 0.5 mL inactive nutrient solution and 4 mL of Rotiszint® eco plus scintillation liquid (Carl Roth, Karlsruhe, Germany). Radioactivity was measured by liquid scintillation counting (Tri-Carb® 2800TR, PerkinElmer Life and Analytical Sciences, Rodgau, Germany). Samples were taken at min 0 (immediately after addition to the plants), 2.5, 5, 7.5, 10, 15, 20, 30, 45, 60 and 75 after addition to the plants, then every 30 min until 5 h, thereafter every hour and after 8 h every 2 h, if needed. The experiment lasted until the LP plants showed no further uptake, but at least 7 h. The experiment was conducted with 6 plants per P nutrient level. The plants were harvested at the end of the experiment and the biomass of the root system was measured.
The uptake rate (Ik,k+1) was calculated between two time points (k and k + 1) with the following equations. Because the measured radioactivity fluctuated strongly in samples of the first minutes of the experiment, only the values after stabilization were used for calculations and models.
$$ {\mathrm{C}}_{\mathrm{k}}={\mathrm{C}}_0\times \frac{{\mathrm{A}}_{\mathrm{k}}}{{\mathrm{A}}_0} $$
Ck: P concentration in the nutrient solution at time point k [μM]
Ak: activity in nutrient solution at time point k [Bq μL−1]
C0: P concentration at time point 0 (start of experiment) [μM]
A0: activity in nutrient solution at time point 0 (start of experiment) [Bq μL−1]
$$ {\mathrm{I}}_{\mathrm{k},\mathrm{k}+1}=\frac{{\mathrm{C}}_{\mathrm{k}+1}\times \mathrm{V}-{\mathrm{C}}_{\mathrm{k}}\times \mathrm{V}}{\frac{{\mathrm{t}}_{\mathrm{k}+1}-{\mathrm{t}}_{\mathrm{k}}}{\mathrm{FM}}} $$
Ik,k+1: P uptake rate between time point k and k + 1 [μmol min−1 g−1]
V: volume of experimental solution [L]
tk, tk+1, : time point k and k + 1 [min]
FM: fresh mass of fine roots [g].
Due to plant uptake, the P concentration in the nutrient solution declined during the experiment. Therefore, the P concentration (Ck,k+1) for each uptake rate (Ik,k+1) was calculated as:
$$ {\mathrm{C}}_{\mathrm{k},\mathrm{k}+1}=\frac{{\mathrm{C}}_{\mathrm{k}}+{\mathrm{C}}_{\mathrm{k}+1}}{2} $$
To model the change in uptake rate in relation to the concentration in the nutrient solution, Ik,k+1 and the corresponding Ck,k+1 were plotted und a curve was fitted to determine vmax (the maximum uptake rate), Cmin (the minimum concentration at which the plants can take up phosphate) and Km (the Michaelis-Menten constant) using the following equation in a non-linear model (R-package “nlme”, [32]):
$$ {\mathrm{I}}_{\mathrm{k},\mathrm{k}+1}=\frac{{\mathrm{v}}_{\max}\times \left({\mathrm{C}}_{\mathrm{k},\mathrm{k}+1}-{\mathrm{C}}_{\min}\right)}{{\mathrm{K}}_{\mathrm{m}}+{\mathrm{C}}_{\mathrm{k},\mathrm{k}+1}-{\mathrm{C}}_{\min }} $$
The fitted curve for these parameters was drawn; the standard errors of the predictions were used for calculation of the 95 %-prediction interval (R-package “emdbook” [33]). Because the uptake rate of the control plants (HP) was slow, a linear fit without slope was defined as vmax. Outliers (>1.5 interquartile ranges below first and above third quartile) were excluded from the linear fit.
RNA extraction and microarray analysis
The first three fully expanded leaves from the top and fine roots (<2 mm diameter) of 59-day-old HP, MP and LP poplars (experiment 2) were harvested and immediately shock frozen in liquid nitrogen. Six plants per treatment (i.e. 18 plants in total) were used and the leaves respective roots of two individual poplars were pooled yielding three biological replicates per treatment. Frozen tissue was milled in liquid nitrogen and RNA was extracted from about 1 g of plant powder according to Chang et al. [34] with modifications: no spermidine was used, 2 % β-mercaptoethanol was added separately from the extraction buffer and phenol:chloroform:isoamyl alcohol (Roti®-Aqua PCI, 25:24:1, Roth, Karlsruhe, Germany) was used instead of chloroform:isoamyl alcohol. RNA concentrations and purity were determined spectrophotometrically via the absorptions at 260 and 280 nm (BioPhotometer, Eppendorf, Hamburg, Germany). RNA integrity was determined with Agilent BioAnalyzer 2100 capillary electrophoresis at the Microarray Facility (MFT Services, Tübingen, Germany). Hybridization on the GeneChip® Poplar Genome Array (Affymetrix, Santa Clara, CA), washing, staining and scanning were conducted at the Microarray Facility (MFT Services, Tübingen, Germany). Raw and normalized data were uploaded into the EMBL-EBI ArrayExpress database [35] under E-MTAB-3934.
Statistical analyses of the raw data were performed as described in Janz et al. [36] using the free statistic software R (version 2.14.2 [37]). Only transcripts that had a detection call of “present” (“mas5calls” function with default settings for tau (0.015), alpha1 (0.04) and alpha2 (0.06)) on all replicate chips of at least one condition were used. For annotation of the microarray probe-set, the best gene model from the annotation file of the Aspen Data Base [38] was used. When several probe sets represented one gene, the mean value of the log2-expression data was used for further analysis. The log2-expression values xi for each gene i of interest were normalized using z-transformation resulting in xi’:
$$ {{\mathrm{x}}_{\mathrm{i}}}^{,}=\frac{{\mathrm{x}}_{\mathrm{i}}-\overline{\mathrm{x}}}{\mathrm{s}}\mathrm{with} $$
\( \overline{\mathrm{x}} \): arithmetic mean value of all log2-expression data for one gene,
s: standard deviation of all log2-expression data for one gene.
A heatmap of normalized expression values (“heatmap.2” function) was created using the R package “gplots” [39].
Here, the members of the putative PHT families 1, 2, 3, and 4 (according to [3, 4]) were retrieved as genes of interest. To obtain the gene IDs and protein sequences for all putative poplar PHT genes, BlastP searches in the database Phytozome v9.1 [40] were conducted. The genome of Arabidopsis thaliana [41], (https://phytozome.jgi.doe.gov/pz/portal.html#!info?alias=Org_Athaliana) was searched with the keyword “pht” extracting the amino acid sequences of eighteen annotated PHT genes. These sequences were used for a BlastP search (e-value cutoff at e-20) against the Arabidopsis genome. The Pfam database was used to identify proteins with functional domains [42]. Additional genes found by the BlastP search with the same domains (Mito_carr, MFS_1, Sugar_tr, Pfam-B_703, PHO4) that are present in the annotated proteins were added to the gene list. These protein sequences were used for BlastP searches against the Populus trichocarpa [43], the Oryza sativa [44], (https://phytozome.jgi.doe.gov/pz/portal.html#!info?alias=Org_Osativa) and the Zea mays genomes [45], (https://phytozome.jgi.doe.gov/pz/portal.html#!info?alias=Org_Zmays). When the resulting proteins had similar domains as Arabidopsis PHTs, they were used for a second BlastP against the respective genome. The resulting sequences were added to the list when the protein domains were similar to the PHT Pfam domains or when the best BlastP hit against the Arabidopsis genome was a PHT.
The phylogenetic tree was created with ClustalW2 and ClustalW2 phylogeny on the EMBL-EBI webpages [46], (http://www.ebi.ac.uk/Tools/msa/clustalw2/ and http://www.ebi.ac.uk/Tools/phylogeny/clustalw2_phylogeny) with default settings and displayed with MEGA6 [47], (http://www.megasoftware.net). Based on the phylogeny, the poplar PHT2 to PHT4 genes were named. Gene IDs of the members of all putative PHTs were compiled in Additional file 1: Table S1. The 1 kb upstream region of each poplar gene coding for a putative phosphate transporter was obtained from Phytozome using the BioMart tool on the Phytozome webpage (https://phytozome.jgi.doe.gov/biomart/martview) and used for a motif search with PLACE [48].
Quantitative Real Time PCR of P transporter genes
RNA samples from the same samples that had been used for microarray analyses (200 ng μL−1 in 25 μL) were treated with Ambion® Turbo DNA-free™ kit (Life Technologies, Carlsbad, CA, USA) two times according to the manual instructions and transcribed to cDNA (0.5 μg) with the RevertAid First Strand cDNA Synthesis Kit and First Strand cDNA Synthesis Kit (Thermo Fisher Scientific, Braunschweig, Germany) using oligo(dT)-primers. For each gene, at least two technical replicates and three biological replicates were analyzed by quantitative Real Time PCR (qRT PCR) with a LightCycler 480® (Roche Diagnostics, Mannheim, Germany). The reaction volume (20 μL) consisted of 10 μL SYBR Green I Master kit (Roche Diagnostics, Mannheim, Germany), 2 μL of the forward and reverse primers (10 μM, Additional file 1: Table S2 for detailed information), 1 μL nucleic free water and 5 μL cDNA-solution (1:10 dilution). After pre-incubation (95 °C, 5 min), 45 or 55 cycles of amplification followed: 95 °C for 10 s, 57 °C (55 °C for PtPHT1;2) for 10 s and 72 °C for 20 s. Melting curve (95 °C for 5 s, 65 °C for 1 min, then to 97 °C at a rate of 0.11 °C s−1) analyses implemented in the LightCycler 480® software were used to assess primer specificity.
To calculate primer efficiency, raw data were converted using LC480 conversion (version 2014.1; www.hartfaalcentrum.nl/index.php?main=files&sub=LC480Conversion) and loaded into LinRegPCR (version 2016.0; [49]). The mean efficiency for each primer pair was calculated over all samples per gene after baseline subtraction. Cq-values were calculated using the fluorescence threshold of 3.597. Relative expression values for each sample were calculated against two reference genes (Potri.001G309500 [Actin], Potri.001G047200 [PPR-repeat gene]):
$$ \mathrm{Relative}\kern3pt \mathrm{Expression}=\frac{\sqrt{{\mathrm{E}}_{\left(\mathrm{R}\mathrm{e}{\mathrm{f}}_1\right)}^{\mathrm{Cq}}\times {\mathrm{E}}_{\left(\mathrm{R}\mathrm{e}{\mathrm{f}}_2\right)}^{\mathrm{Cq}}}}{{\mathrm{E}}_{\left(\mathrm{G}\mathrm{O}\mathrm{I}\right)}^{\mathrm{Cq}}} $$
E: efficiency of primer for gene
Cq: quantification cycle value of sample for gene
(GOI): gene of interest
(Refi): reference gene i [50].
Statistical analyses
R (versions 2.14.2 and 3.0.2; [37]) was used for all statistical analyses. Mean value ± standard error were calculated. One- or Two-Way-ANOVA and Tukey’s HSD were performed on original or transformed data. Residuals were tested visually for normal distribution and homogeneity of variance. Data were transformed logarithmically (log2) or by square root, if needed. For statistical comparisons of single kinetic parameters, Welsh’s t-test was performed using the output data of the models. For percentage data on 33P recovery and P allocation, a general linear model with binomial distribution was fitted on underlying count data, and an analysis of deviance was used to calculate significant factor and interaction effects. A subsequent Tukey test was performed to determine the homogenous subsets. Means were considered to differ significantly between treatments, if p ≤ 0.05. Differences between treatments are shown in figures and tables with different letters. The p-values calculated for gene expression data and vmax for kinetic data were adjusted by Bonferroni correction. Two-Way-Repeated measurement-ANOVA with Tukey’s HSD was performed for plant growth over time.