Introduction

Phosphorus (P), taken up as inorganic phosphate (Pi), is a plant macronutrient and therefore, essential for plant productivity. Primary roles for P within the plant include metabolic energy transfer, such as in photosynthesis and respiration, along with being an integral component of nucleic acids and phospholipids. Therefore, to ensure that P is not limiting to crop growth, large quantities of P fertiliser are applied to agricultural systems. High rates of P fertiliser application are a concern, because global phosphate rock reserves are finite, and because excess Pi leads to ecosystem eutrophication (Blackwell et al. 2019; Fixen and Johnston 2012). Much research has focussed on the molecular responses of plants to variation in Pi supply. These findings have elucidated a complex network of response patterns to sense, signal and transport Pi (Chiou and Lin 2011; Puga et al. 2017; Wang et al. 2018; Yang and Finnegan 2010). The movement of P throughout a plant involves several families of Pi transport proteins (Srivastava et al. 2018). Members of the PHOSPHATE TRANSPORTER1 (PHT1) family of H+ / Pi symporters are located on the plasma membrane and are responsible for the import of Pi into cells throughout the plant (Victor Roch et al. 2019). In roots, these transporters are located at the root-soil interface and are responsible for Pi uptake (Karthikeyan et al. 2002; Misson et al. 2004; Mudge et al. 2002). The expression of these proteins is tightly regulated at the transcriptional level, and at a number of post-translational levels, particularly in response to Pi (Bayle et al. 2011).

There are nine PHT1 gene family members in Arabidopsis thaliana (Mudge et al. 2002). Expression analyses of the various tissues of A. thaliana have allowed the likely location of the different AtPHT1 proteins to be mapped and putative roles identified (Nussaume et al. 2011). Nearly all of the AtPHT1 genes are repressed by Pi (Lapis-Gaza et al. 2014; Misson et al. 2005). Studies examining Pi responsiveness of PHT1 gene expression in plants such as A. thaliana and crops such as Oryza sativa (rice), Zea mays (maize), Setaria italica (foxtail millet) and Triticum aestivum (wheat) often include a high-P treatment ranging from 200 µM to 500 µM Pi (Ai et al. 2009; Ceasar et al. 2014; Misson et al. 2004; Nagy et al. 2006; Teng et al. 2017). These Pi supplies are over 20-fold higher than would occur in natural soil solutions, with most agricultural soil solutions only containing up to 11 µM Pi (Lambers 2022). Studies on PHT1 expression have often taken the perspective of the potential induction of PHT1 expression under low-P or P-starved conditions. However, high Pi supplies can be toxic, leading many plant species to reduce Pi uptake by repressing transporters once excess Pi is reached in high-P conditions (Lambers 2022). Therefore, this study takes an alternative perspective, asking whether a P-sensitive species from a severely P-impoverished environment has lost the capacity to down-regulate PHT1 expression at elevated Pi supplies, leading to its high P sensitivity.

Hakea prostrata (Proteaceae) is a P-sensitive species widely distributed in southwestern Australia, which has some of the most severely P-impoverished soils in the world (Kooyman et al. 2017). These ancient and highly-weathered soils support a global biodiversity hotspot for plants, with the scarceness of plant-available Pi a likely driver for this biodiversity (Hayes et al. 2021; Hopper and Gioia 2004). Proteaceae are a prominent family in this region and exhibit a range of P-efficiency traits (Hayes et al. 2021). One adaptation shared amongst almost all Proteaceae is the ability to form cluster roots, greatly enhancing P acquisition from these ancient soils (Lambers et al. 2008). Cluster roots have a brush-like appearance and occur on lateral roots where a region of compact ‘rootlets’ extends out from the growing root axis (Fig. 1) (Shane and Lambers 2005). Hakea prostrata cluster roots can have over 1000 rootlets per 10 mm and are mature at around 12 to 13 days after emergence, after which they begin to senesce (Lamont 2003; Shane et al. 2004a). The mature cluster root phase of H. prostrata coincides with a large exudative burst of the carboxylates malate and citrate (Shane et al. 2004a). Another root trait appearing among many cluster-rooted species is the lack of a complete exodermis (Lambers et al. 2018). This trait is likely important for the massive release of carboxylates, and subsequent nutrient acquisition (Lambers et al. 2018). In H. prostrata, increasing the external P supply systemically suppresses the formation of cluster roots, supporting their primary function of ‘mining’ soil-sorbed P (Shane et al. 2004c).

While Proteaceae such as H. prostrata are generally highly effective at mining P from impoverished soils, they can also exhibit foliar P toxicity symptoms, even under relatively low soil P concentrations (Shane et al. 2004b). For example, foliar P toxicity symptoms develop in six-month-old Banksia ericifolia (Proteaceae) seedlings supplied with 60 mg P L−1 potting mix (Parks et al. 2000) and five-month-old Banksia grandis seedlings supplied with 25 µg P g−1 nutrient-poor sand (Lambers et al. 2002). In hydroponically-grown two-month-old H. prostrata seedlings, foliar toxicity symptoms begin to appear when plants are supplied with above 100 µmoles P per day (Shane et al. 2004c). A low capacity to down-regulate uptake of P in the roots has been identified in H. prostrata as an underlying cause of P sensitivity (Shane et al. 2004c). This inability to regulate P uptake was also demonstrated for other P-sensitive species, such as in seedlings of five-month-old Banksia attenuata and Banksia menziesii that have evolved in similarly P-impoverished environments to that of H. prostrata (de Campos et al. 2013). Both Banksia species exhibit P toxicity symptoms when supplied with 20 µmoles P per day (de Campos et al. 2013). However, not all Proteaceae are P-sensitive when grown under high-P conditions. For example, Grevillia crithmifolia (Proteaceae) maintains sub-toxic foliar P concentrations at high external P supply (Shane and Lambers 2006). Two contributing factors for lower foliar P concentrations in G. crithmifolia are a greater ability to down-regulate P uptake and a greater capacity to store excess P in root tissues (Shane and Lambers 2006). In addition to the inability to down-regulate Pi uptake, H. prostrata and several other Proteaceae also preferentially allocate P to photosynthetic cells (Hayes et al. 2018; Shane et al. 2004b). This trait ensures high photosynthetic P-use efficiency, but comes at the cost of P accumulation in photosynthetically-active mesophyll cells under elevated P supplies rather than allocation of excess P to epidermal cells, as occurs in other species (Shane et al. 2004b).

In this study, we used a recent H. prostrata transcriptome assembly (Bird et al. 2024) to identify HpPHT1 transcript sequences and investigated their relative expression in both newly-emerged white roots and mature cluster roots in response to differences in external Pi supply. We hypothesised that HpPHT1 expression in both white roots and cluster roots would not differ considerably under low and high Pi supplies, thereby identifying a mechanism that contributes to the inability of this species to regulate P uptake. The findings of this study will further our understanding of the poor ability of H. prostrata and other Proteaceae to regulate P-uptake, and therefore, avoid P toxicity.

Materials and methods

Sequence identification, analysis and primer design

A recently-assembled transcriptome of H. prostrata was used to identify HpPHT1 transcripts (Bird et al. 2024) (Transcriptome Shotgun Assembly accession GKDQ01000000). The transcriptome was generated using RNA sequences from white roots and mature leaves from plants grown in hydroponics at three Pi supplies, four developmental stages of cluster roots from plants grown in hydroponics, and five developmental stages of leaves from plants growing in their natural habitat. A total of 53,710 putative peptide sequences were encoded by this transcriptome, which corresponded to 51,710 representative transcripts. The two most divergent AtPHT1 sequences, AtPHT1;1 and AtPHT1;9 (The Arabidopsis Information Resource (Berardini et al. 2015) AT5G43350, AT1G76430), were each used as query nucleotide sequences to retrieve homologous H. prostrata transcript sequences using the Basic Local Alignment Search Tool (BLAST) with a maximum E-value of 1e−5 (Geneious, Biomatters, Auckland, New Zealand). The two retrieved sequence sets were combined and duplicates removed. To confirm these sequences as putative PHT1 sequences, each sequence was used as a query sequence in reverse BLAST searches against the NCBI nucleotide database (National Centre for Biotechnology Information, Bethesda, MD, USA). The full-length AtPHT1;1 open reading frame (ORF) is 1,572 bases, so HpPHT1 transcript sequences encoding an ORF less than 800 bases in length were removed from further analysis. The RNA reads used to produce the transcriptome came from multiple individuals, so allelic differences were possibly present. Therefore, multiple and pairwise alignments were used to identify probable allelic sequences. Sequences with greater than 95% nucleotide identity were considered allelic and removed.

Primer pairs for quantitative real-time PCR (qPCR) were designed (Primer3 v2.3.7, SourceForge, San Diego, CA, USA in Geneious, Biomatters, Auckland, New Zealand) to anneal near the 3’ end of its target transcript and to amplify the region approximately 200 bp upstream to 250 bp downstream of the stop codon of the ORF. Default design parameters were used, which included a Tm optimum of 60 °C and a GC content of 50%. Primer sequences were visually cross-checked to ensure unique sequences were targeted. Primers were synthesised by Thermo Fisher Scientific, Waltham, MA, USA (Table S1). The peptide sequences deduced from the identified HpPHT1 transcripts were compared in a relational tree (Geneious tree builder with default settings, Geneious, Biomatters, Auckland, New Zealand). The tree included the nine well-characterised AtPHT1 protein sequences, as well as PHT1 protein sequences extracted from NCBI RefSeq (O’Leary et al. 2016), including predicted protein sequences of Macadamia integrifolia (Proteaceae, GCF_013358625.1), Telopea speciosissima (Proteaceae, GCF_018873765.1) and Nelumbo nucifera (Nelumbonaceae, GCF_000365185.1). These PHT1 protein sequences were retrieved by BLAST (blastp) with a maximum E-value of 1e−20 using the nine AtPHT1 protein sequences as query sequences (The Arabidopsis Information Resource (Berardini et al. 2015) AT5G43350, AT5G43370, AT5G43360, AT2G38940, AT2G32830, AT5G43340, AT3G54700, AT1G20860, AT1G76430). Genetic distances amongst all sequences were calculated using the Jukes-Cantor model (Jukes and Cantor 1969). The translated HpPHT1 peptide sequences were aligned using the blosum62 cost matrix (Henikoff and Henikoff 1992) (Geneious, Biomatters, Auckland, New Zealand).

Plant growth and measurements

Hakea prostrata seeds (Nindethana Seed Service, Albany, WA, Australia) were surface sterilised in 5% (w/v) hypochlorite and germinated on MilliQ H2O-moistened filter papers in Petri dishes at 15 °C with a 12 h light-dark cycle. The growth schedule detailed in Supplementary Methods was followed until seedlings were approximately 15 weeks old. At this point, individual plants of a similar size were grown in individual 4.5 L pots, containing 4 L of continuously-aerated low-nutrient solution (Shane et al. 2003), which was replaced three times per week. Plants were assigned to one of four P (as KH2PO4) treatments with supplies of 0.1, 0.8, 6 and 12 µM P. These treatments were equivalent to supply rates of 0.17, 1.35, 10 and 20 µmoles P per day. The P concentration was increased weekly for the higher treatments of 6 and 12 µM P until the final supply rates were reached after four weeks. Incremental increases allowed acclimation to the new P supply and minimised the potential for a shock response to a suddenly higher P supply. The final treatment of 0.8 µM P was increased to this level from 0.4 µM P after six weeks based on a lack of observable differences in growth between the 0.1 µM and 0.4 µM P supplies.

Plants were all harvested at approximately 7.5 months of age and all treatments had been at their final P supply for at least seven weeks. Young white roots (0 to 50 mm zone of white growth from root tips), mature leaves (fully expanded and dark green) and mature cluster roots (Fig. 1) were harvested, and leaves and cluster roots were counted. A representative portion of each organ type from each plant was weighed before snap freezing in liquid nitrogen and stored at -80 °C. The remaining part of each organ type was weighed and oven-dried at 60 °C for one week before determining dry weights separately.

Root samples for surface area determinations were harvested from a separate group of plants as described in Supplementary Methods. For surface area determination, six replicates for both mature cluster roots and pooled white root samples (0 to 100 mm zone of white growth from root tips) were harvested, blotted dry with paper towels and weighed to determine their fresh weight (FW). Cluster-root samples were then dissected to separate rootlets from the lateral root before all samples were scanned at 600 dpi and the total surface area measured (WinRHIZO 2009, Regent Instruments Inc., Québec, Canada).

Fig. 1
figure 1

Four-month-old Hakea prostrata plant producing cluster roots (white arrow) in aerated hydroponics one week after treatment commenced (see Methods). a) H. prostrata plant in hydroponics and b) Mature cluster roots, approximately 12 to 13 days after emergence

Organ phosphorus concentrations

Approximately 30 mg of each dried organ sample was transferred to a 25-ml flask for acid digestion to convert esterified P to Pi (Zasoski and Burau 1977). In summary, samples were digested in 3 ml of 70% (v/v) HNO3 at 90 °C to 100 °C, followed by digestion in 0.5 ml of 70% (v/v) HClO4 at 150 °C to near dryness. After samples had cooled, 1 ml of 32% (v/v) HCl was added before heating samples at 130 °C for a further 5 min. After cooling, deionised water was added to bring each sample to a final volume of 10 ml. Total Pi concentration was measured colorimetrically using a malachite green-based method (Motomizu et al. 1983). Four to five replicates for both white roots and leaves, and three to four replicates for cluster roots were measured.

RNA extraction, reverse transcription and quantitative PCR

RNA was extracted from snap-frozen root and cluster root samples using a commercial kit following the manufacturer’s instructions (Spectrum™ Plant Total RNA Kit, Merck, Darmstadt, Germany). Extracted RNA was DNase treated following the manufacturer’s instructions (RQ1 RNase-Free DNase, Promega, Madison, WI, USA). DNase-treated RNA was purified once more by extracting with 24:1 chloroform : isoamyl alcohol and precipitated by adding 1/10 volume 3 M sodium acetate, pH 5.6 and 1 volume isopropanol. The RNA precipitate was collected by centrifugation, washed with 70% (v/v) ethanol, air-dried and resuspended in nuclease-free water. Total RNA concentration was measured spectrophotometrically (Nanodrop 1000 spectrophotometer, Thermo Fisher Scientific, Waltham, MA, USA) and up to 1 µg RNA was reverse-transcribed using a commercial kit following the manufacturer’s instructions (Tetro™ cDNA Synthesis Kit, Bioline, London, UK).

Each quantitative PCR (qPCR) assay contained 2.5 µL cDNA template (25 ng cDNA), 2.5 µL of the target-specific primer mix (0.6 µM per primer) and 5 µL low-ROX SYBR green dye reaction mix (KiCqStart™, Sigma-Aldrich, St. Louis, MO, USA). Real-time qPCR was carried out on a thermocycler with 40 cycles of denaturation (3 s at 95ºC) followed by primer annealing and extension (30 s at 65ºC) (7500 FAST Real-Time PCR System; Applied Biosystems, Waltham, MA, USA). Three biological replicates, each with three technical replicates, were measured for all conditions tested. Relative transcript expression levels were normalised to the housekeeping genes ACTIN7 (Hong et al. 2008) and YLS8 (Czechowski et al. 2005), which have been validated for H. prostrata (Kuppusamy et al. 2014). Cycle thresholds for product amplification are reported on a log2 scale, with the mean for each target sequence reported as a 40-∆Ct value (Kuppusamy et al. 2014).

Statistics

Statistics were performed in R software using one-way ANOVA followed by a Tukey post hoc test for pairwise comparisons in the ‘lsmeans’ package (Lenth 2016; R Core Team 2022).

Results

Sequence identification and analysis

Ten PHT1 transcript sequences were identified within a previously published H. prostrata transcriptome constructed using RNA sequences from white roots, cluster roots and leaves. Seven of these transcripts encoded a full-length PHT1 ORF (Fig. S1). Full-length ORFs ranged in length from 525 amino acids (AA) for HpPHT1;1 to 544 AA for HpPHT1;3 (Fig. S1). Incomplete transcripts were found for HpPHT1;2, HpPHT1;4 and HpPHT1;7, which encoded partial ORFs 409, 306, and 412 AA in length, respectively. Incomplete transcripts were all truncated at the 5’ end and extended beyond the stop codon into the 3’ untranslated region (Fig. S1). Deduced peptide sequences from all 10 HpPHT1 transcript sequences, except the truncated HpPHT1;4 sequence, contained the PHT1 signature motif GGDYPLSATIxSE, a domain conserved across PHT1 protein sequences from a large range of plant species (Karandashov and Bucher 2005) (Fig. S1). The motif was likely missing from HpPHT1;4 only because the partial transcript did not extend far enough toward the 5’ end.

Peptide sequences HpPHT1;1, HpPHT1;3, HpPHT1;5, HpPHT1;6 and HpPHT1;10 all had approximately 92% amino acid similarity to the AtPHT1;1 protein sequence. Sequences that were most divergent from AtPHT1;1 were HpPHT1;8 and HpPHT1;9, both with approximately 80% similarity (Table S2). Among the seven complete HpPHT1 sequences, the highest similarity was 96% between HpPHT1;1 and HpPHT1;3, while the lowest similarity was about 77% between HpPHT1;8 and both HpPHT1;5 and HpPHT1;10 (Table S2).

The peptide sequences of all 10 HpPHT1 ORFs were compared with PHT1 ORFs from four other species in a phylogenetic tree (Fig. 2). This comparison included the nine sequences of the well-characterised AtPHT1 family, five sequences from the basal dicot N. nucifera (Nelumbonaceae), 14 sequences from M. integrifolia (Proteaceae) and 15 sequences from T. speciosissima (Proteaceae). The phylogenetic tree had three main clades. The most divergent clade, Clade I, included HpPHT1;8 and HpPHT1;9 and two sequences from each Proteaceae, as well as AtPHT1;8 and AtPHT1;9 (Fig. 2). Interestingly, there was only one sequence from N. nucifera in this clade. Another clade, Clade II, contained only sequences from the Proteaceae and included HpPHT1;2, HpPHT1;4, HpPHT1;7 and HpPHT1;10 (Fig. 2). In addition to the four H. prostrata sequences, there were six sequences from both M. integrifolia and T. speciosissima. The final main clade, Clade III, contained HpPHT1;1, HpPHT1;3, HpPHT1;5 and HpPHT1;6, and six AtPHT1 sequences (Fig. 2). Within this clade, the AtPHT1;5, AtPHT1;1, AtPHT1;2 and AtPHT1;3 sequences were not closely related to the HpPHT1 sequences. No Proteaceae sequences grouped closely with AtPHT1;6 and this sequence did not group within the three main clades.

Fig. 2
figure 2

Unrooted phylogenetic tree showing predicted amino-acid sequence relatedness among 10 Hakea prostrata PHT1 protein sequences (blue text), nine Arabidopsis thaliana PHT1 protein sequences (red text), and identified PHT1 protein sequences from the basal dicot Nelumbo nucifera (Nelumbonaceae, purple dots), Macadamia integrifolia (Proteaceae, green dots) and Telopea speciosissima (Proteaceae, light blue dots). The scale bar shows the relative genetic distance calculated using the Jukes-Cantor method (Jukes and Cantor 1969)

Response of growth to phosphorus supply

Treatments were designed to emulate conditions of P-impoverished (0.1 µM Pi), low-P (0.8 µM Pi), elevated-P (6 µM Pi), and excessive-P (12 µM Pi) supplies relative to the P-impoverished natural habitat of H. prostrata. For simplicity of presentation, the treatments are designated P0.1, P0.8, P6 and P12, respectively, in the text. These supplies led to clear morphological differences in plant growth (Fig. 3). Shoot biomass was three-fold greater at P6 and P12 than at P0.1 and P0.8 (Fig. 4a), while growth at P0.1 was indistinguishable from that at P0.8, and growth at P12 was indistinguishable from that at P6. The differences in shoot biomass were reflected in average leaf number (Fig. 4b). The average number of leaves was 14 at P0.1 and P0.8 and 40 at P6 and P12 (Fig. 4b). Non-cluster root biomass showed a similar trend to that of the shoots. Non-cluster root production was 2.5- to 3-fold greater at P6 and P12 than at P0.1 and P0.8 (P < 0.05) (Fig. 4a). Cluster root formation was suppressed by higher Pi supply, especially comparing P0.8 with P12 (P < 0.05) (Fig. 4a, b). The suppression was reflected in both the average dry weight (DW) and the number of cluster roots per plant (Fig. 4a, b). Cluster root abundance was highest at P0.8 in this experiment (Fig. 4b). Mature cluster roots had almost twice the surface area of white roots on a FW basis (Fig S2), highlighting the relative absorptive capacity of cluster roots and white roots.

Fig. 3
figure 3

Representative Hakea prostrata plants grown in an aerated hydroponic system at phosphate supplies of 0.1, 0.8, 6, and 12 µM Pi. Nutrient solutions were changed three times per week. Cluster roots are indicated with white arrows. Photos: Hongtao Zhao

Fig. 4
figure 4

Root and leaf mass of Hakea prostrata plants grown in an aerated hydroponic system at different phosphorus (P) supplies. a) Dry mass of non-cluster roots (blue), cluster roots (red) and shoots (green). b) Mean counts of the number of cluster roots and leaves. Means ± standard error (n = 15) are shown. Different letters denote significant differences among treatments and for an organ (P < 0.05)

Root mass ratio (RMR) showed that the overall root-system size responded negatively to higher Pi supply from P0.1 to P6 (P < 0.05) (Fig. 5). This significant size decrease was almost entirely due to lower investment in the formation of cluster roots, as the RMR for the non-cluster root fraction was around 0.3 at all Pi supplies (Fig. 5). Notably, neither the cluster root nor the non-cluster root fractions differed in RMR between P6 and P12.

Fig. 5
figure 5

Root dry mass ratios for Hakea prostrata grown in hydroponics at phosphorus (P) supplies of 0.1, 0.8, 6, and 12 µM P. The proportion of total plant mass allocated to the whole root systems, and to the cluster root and non-cluster root fractions of the whole roots are shown as means ± standard error (n = 15). Different letters denote significant differences within the whole root system or its components (P < 0.05)

Response of organ P concentrations to external P supply

Mature leaves, mature cluster roots, and white roots all had a significantly higher P concentration at higher Pi supplies than at lower Pi supplies (P < 0.05) (Fig. 6). The P concentration in white roots was higher at each progressively higher P supply. The P concentrations in cluster roots tended to be higher for each progressively higher Pi supply. However, this difference in concentration was only statistically significant at P12. The P concentration in leaves was similarly low at 0.17 mg P g−1 DW at P0.1 and 0.45 mg P g−1 DW at P0.8, but was significantly higher at P6. The P concentration of all organs, except cluster roots, was higher at P12 than at P6 (Fig. 6). Among the organs, mature leaves had the lowest P concentration at P0.1, P0.8 and P6. Mature cluster roots had a higher P concentration than white roots at P0.8 and P6 (P < 0.05) (Fig. 6). The P concentration was indistinguishable among the organs at P12 (P < 0.05) (Fig. 6). Given the scleromorphic nature of Proteaceae leaves, FW foliar P concentrations were also calculated based on Kuppusamy (2018) at a DW : FW ratio of 0.45 (Fig. S3).

Fig. 6
figure 6

Phosphorus (P) concentrations on a dry mass basis in mature leaves, mature cluster roots and white roots of Hakea prostrata grown in aerated hydroponics with different external P supplies. Bars show the means ± standard errors (n = 4 or 5, except n = 3 for cluster roots of plants grown at 6 and 12 µM Pi). Different uppercase letters denote significant differences within an organ among treatments (P < 0.05). Different lowercase letters denote significant differences among organs at the same P supply (P < 0.05)

Relative PHT1 transcript abundance

The relative mRNA abundance for eight HpPHT1 sequences was determined in white roots and cluster roots of plants grown at various Pi supplies (Fig. 7). HpPHT1;5 and HpPHT1;10 were excluded from the analysis after preliminary qPCR showed they had no (or a very low) signal (Ct > 36 cycles) in white roots. For simplicity, differences in transcript quantities for a gene are given as the ratio between the lower P supply and the higher P supply or vice versa, with positive (+) values designating a repression by P and negative (-) indicating an induction by P. In white roots, all eight HpPHT1 genes tested had their highest relative expression at P0.1 (P < 0.05) (Fig. 7a). Transcripts from HpPHT1;1, HpPHT1;3, HpPHT1;6, and HpPHT1;7 were the most abundant and statistically indistinguishable from each other (P < 0.05) (Fig. 7a). The expression of HpPHT1;6 was not responsive to Pi, while the other three highly expressed genes were variously repressed by higher Pi supplies. However, none of these highly expressed genes had a P0.1/P12 greater than 5 in the white roots (P < 0.05) (Fig. 7a). The level of repression did not significantly increase at P supplies above P0.8 for HpPHT1;3 and HpPHT1;7, or above P6 for HpPHT1;1 (P < 0.05) (Fig. 7a). Two other HpPHT1 genes that were repressed by Pi in white roots were HpPHT1;2 and HpPHT1;9. Transcripts from these genes were low in abundance. At P0.1, transcripts from HpPHT1;2 were less than 10% of the abundance of transcripts from each of the four most prominent genes, whilst HpPHT1;9 was less than 20% of the abundance. However, HpPHT1;2 was the most strongly repressed HpPHT1 gene by Pi in white roots, with a P0.1/P12 of about 300. In contrast, HpPHT1;9 had a very low level of repression in response to Pi with a P0.1/P12 of 3 (P < 0.05) (Fig. 7a). Expression of HpPHT1;4 and HpPHT1;8 did not differ among the P treatments in white roots, similarly to the highly expressed HpPHT1;6.

Fig. 7
figure 7

Relative expression of eight Hakea prostrata PHT1 genes in response to different phosphorus (P) supplies. Relative expression of HpPHT1 genes was determined for a) white roots and b) mature cluster roots. Means of relative expression are shown as 40-∆Ct, which is a log2 scale ± standard error (n = 3). 40-∆Ct was calculated relative to the average transcript abundance of a set of reference genes (see Methods). Different letters denote significant differences within an organ for each specific target from plants grown at 0.1, 0.8, 6, or 12 µM P (P < 0.05). * denotes the statistically-indistinguishable most abundant HpPHT1 transcripts (P > 0.05) for both white roots (a) and cluster roots (b). Ratios given in the text to describe differences in transcript abundance between treatments are denoted by a plus sign (+) for higher abundance or a minus sign (-) for lower abundance. Ratios given in the text are in base 10, following conversion from the log2 values shown in this figure

The HpPHT1 transcript patterns within mature cluster roots were generally similar to those in white roots with several notable differences. Transcripts from all genes had the highest or equally highest abundance at P0.1 (P < 0.05) (Fig. 7b). The most abundant PHT1 transcripts within mature cluster roots were from HpPHT1;6 and HpPHT1;7. Relative expression of HpPHT1;6 was not responsive to P in mature cluster roots, similar to the situation in white roots. On the other hand, HpPHT1;7 was repressed by Pi in mature cluster roots, with a P0.1/P6 of 4 (P < 0.05) (Fig. 7b). The next most abundant PHT1 transcripts were from HpPHT1;1, HpPHT1;2 and HpPHT1;3. HpPHT1;1 did not respond to Pi in mature cluster roots, unlike the small repression seen in white roots (P0.1/P12 = 3.7) (Fig. 7a). In contrast to white roots, HpPHT1;2 transcripts were one of the most predominant HpPHT1 transcripts in mature cluster roots (Fig. 7b). Furthermore, HpPHT1;2 and HpPHT1;3 were the most strongly Pi-repressed HpPHT1 genes in cluster roots. HpPHT1;2 transcripts had a P0.1/P0.8 of 4.5, with further repression at higher Pi (P0.8/P6 = 4.3) (P < 0.05) (Fig. 7b). Transcripts for HpPHT1;3 were repressed to a similar extent as those from HpPHT1;2, but only when the Pi supply was above P0.8. Similarly to the findings for white roots, HpPHT1;4 and HpPHT1;8 expression did not significantly differ among the P treatments in mature cluster roots, although HpPHT1;8 did trend toward being induced by higher Pi supplies (P12/P0.1 of -4) (P < 0.07) (Fig. 7b) and was the only HpPHT1 gene to show any signs of induction by Pi. HpPHT1;9 was unresponsive to Pi in cluster roots, in contrast to its repression in white roots.

Discussion

Physiological relevance of hydroponic P supplies for Hakea prostrata

Phosphorus toxicity in H. prostrata and other Proteaceae has been described in several growth experiments (Hawkins et al. 2008; Hayes et al. 2019; Shane et al. 2004c). This toxicity is at least partly explained by the low capacity of H. prostrata to regulate P uptake (Shane et al. 2004c). Therefore, the present study took a different perspective to the majority of studies that have investigated P responses in plants. Previous studies on typical P-demanding plants, such as crop species, often focussed on the P-starvation response and the induction of P uptake systems by low P supplies. Instead, we investigated whether the P-sensitive H. prostrata can down-regulate the expression of transporter genes responsible for Pi uptake to the same extent as other plant species when exposed to elevated Pi supplies. The P treatments used in this study were based on previous hydroponic work on H. prostrata that investigated the basis of P toxicity in this species (Shane et al. 2003, 2004b, c). The lowest two Pi treatments used in the present study were designed to reflect the P-impoverished natural habitat of H. prostrata. The P0.8 treatment gave a mean P concentration in mature leaves of 0.45 mg g−1 DW, which is near the top of the range of approximately 0.2 and 0.5 mg P g−1 leaf DW observed in H. prostrata growing in its natural habitat (Lambers et al. 2012; Sulpice et al. 2014; Yan et al. 2019), while the leaf P concentration in P0.1 (0.17 mg P g−1 leaf DW) was at the bottom of the range. Mature leaf P concentration in P6, at 2.4 mg P g−1 leaf DW, was about five-fold above the highest leaf P concentrations recorded for H. prostrata growing in its natural habitat. Thus, a supply of 6 µM Pi evidently surpassed the capacity of the root and stem tissues to store P and prevent transport and accumulation of excess P in leaves. Preferential P allocation, regarding different organs, was further demonstrated by the increase in P concentration in roots between the lowest treatments, whilst leaf concentration was constant. The mature leaf P concentration of about 6.6 mg P g−1 leaf DW when plants were exposed to excessive P in P12 was not high enough to induce visible foliar P toxicity symptoms. Previously, P toxicity symptoms were observed in mature H. prostrata leaves containing around 9 mg P g−1 DW (Shane et al. 2004b, c). In another study on the impact of Ca2+ on P relations in several Proteaceae, P toxicity symptoms in mature H. prostrata leaves were observed in hydroponically-grown plants supplied with 17 µmoles P per day (Hayes et al. 2019), similar to P12 (20 µmoles P per day). Leaf P concentration in the Ca2+ study was approximately 5 mg P g−1 DW. However, these plants were also given a much higher Ca2+ supply than those in the present study, a treatment that enhances P sensitivity (Hayes et al. 2019). Thus, the leaf P levels in P12 here are sufficient to cause P toxicity in H. prostrata under certain conditions, even though we did not see toxicity symptoms in our plants.

PHT1 gene family expression in Hakea prostrata roots

The identification of 10 representative PHT1 genes in the H. prostrata transcriptome is typical of the total number of PHT1 genes found among examined dicots (Nussaume et al. 2011). However, this is a conservative estimate of the actual number of PHT1 genes in H. prostrata. On one hand, the H. prostrata transcriptome did not represent all plant organs or developmental stages, notably lacking transcripts specific to flowers, fruits, and stems. On the other hand, the high level of sequence conservation found in this gene family in other species (Nussaume et al. 2011) gives the possibility that some highly identical PHT1 transcripts may have been collapsed together or disregarded as alleles during transcriptome assembly. In A. thaliana, for example, AtPHT1;1 and AtPHT1;2 have approximately 98% nucleotide sequence identity (Mudge et al. 2002). The actual number of HpPHT1 genes may be closer to the 14 and 15 PHT1 sequences found in the genome assemblies of the Proteaceae M. integrifolia and T. speciosissima, respectively (Chen et al. 2022; Nock et al. 2020).

Within H. prostrata, the proliferation of non-cluster roots in the P6 and P12 treatments compared with the ecologically-relevant P0.1 and P0.8 treatments was an important aspect of this study, even though the overall DW and therefore absorptive surface area of cluster roots that were produced was only significantly lower in the P12 treatment. The observed lower relative abundance of cluster roots in P6 indicated a shift in root physiology, because P-uptake rates by H. prostrata only begin to decline once the relative abundance of cluster roots is at least partly suppressed (Shane et al. 2003). In H. prostrata cluster roots, the prominent transcripts, HpPHT1;2 and HpPHT1;3, were suppressed in response to a higher Pi supply and were more strongly down-regulated than other HpPHT1 genes. HpPHT1;2 responded to Pi between the habitat-relevant P0.1 and P0.8 treatments with a P0.1/P0.8 of 4.5. This Pi repression was similarly strong at P6 (P0.8/P6 = 4.3). In contrast, HpPHT1;3 expression was only significantly repressed compared with P0.1 in the P6 and P12 treatments (P0.8/P6 = 8), above the ecologically-relevant treatment range. The prominence and repression of HpPHT1;2 by Pi within ecologically-relevant P supplies suggests that this gene has a specialised role in cluster roots. In this light, it is interesting that the protein encoded by HpPHT1;2 falls within the Proteaceae-specific clade of PHT1 protein sequences, along with HpPHT1;4 and HpPHT1;7.

Potential mechanisms contributing to P sensitivity in Hakea prostrata

The traits contributing to P sensitivity in H. prostrata are likely to reside in other parts of the root outside ephemeral cluster roots, the formation of which is repressed by high Pi concentrations. Potential genetic mechanisms contributing to P sensitivity in this species include the number of highly-expressed HpPHT1 genes in white roots and the relative insensitivity of some of these genes to repression by Pi. The relatively high levels of expression of HpPHT1;1, HpPHT1;3, HpPHT1;6 and HpPHT1;7 in white roots suggested that they encode the main proteins involved in Pi uptake from the external medium. The predominant PHT1 transcripts in A. thaliana roots are from AtPHT1;1, AtPHT1;2 and AtPHT1;4 (Table S3) (Lapis-Gaza et al. 2014; Morcuende et al. 2007; Scheible et al. 2023). Thus, H. prostrata has an additional PHT1 gene that is highly expressed in roots compared with that in A. thaliana, which would allow a higher influx of Pi when supplies are elevated.

The abundance of several HpPHT1 transcripts was repressed by Pi, as is generally the case for PHT1 transcripts in plant species (Lambers 2022; Lapis-Gaza et al. 2014). However, this repression is likely attenuated in H. prostrata in comparison with other species. Prominently-expressed HpPHT1;1, HpPHT1;3 and HpPHT1;7 responded to Pi similarly to two of the three most-predominant PHT1 transcripts in A. thaliana, AtPHT1;1 and AtPHT1;2 (Table S3). In white roots of H. prostrata, the three HpPHT1 genes were repressed by Pi with a P0.1/P12 of about 4 to 4.5. HpPHT1;3 and HpPHT1;7 were maximally repressed in the ecologically-relevant treatment P0.8, while HpPHT1;1 was maximally repressed only at ecologically less-relevant treatments P6 and P12. Despite having three members that are moderately responsive to Pi supply, the HpPHT1 family seems to entirely lack a highly-expressed member that is as strongly Pi-responsive as the highly-expressed AtPHT1;4. This A. thaliana gene is typically expressed at least 20-fold more abundantly at low P supplies than at high P supplies (Table S3) (Barragán-Rosillo et al. 2021; Bustos et al. 2010; Lapis-Gaza et al. 2014; Misson et al. 2005; Morcuende et al. 2007; Scheible et al. 2023). Moreover, the final of the four highly-expressed PHT1 genes in H. prostrata, HpPHT1;6, was completely insensitive to Pi supply and was maximally expressed even at P12. Therefore, the combination of one highly-expressed HpPHT1 gene that is less responsive to Pi and one highly-expressed HpPHT1 that is unresponsive to Pi has the potential to increase constitutive Pi uptake at the soil interface at elevated Pi supplies and contribute to the increased P sensitivity of this species (Shane et al., 2004a).

The assignment of potential roles to highly-expressed HpPHT1 genes was greatly aided by the detailed knowledge available from studies of A. thaliana. These studies have provided a comprehensive AtPHT1 gene list supported by extensive work on the expression and function of many AtPHT1 gene family members. While work in other species is not as extensive, it is still informative with respect to Pi transport in non-model plant species such as H. prostrata. In the woody dicot Populus x canescens, PcPHT1;1 and PcPHT1;2, which are orthologous to AtPHT1;1 and AtPHT1;2, were highly repressed by Pi, having PL/PH ratios of 61 and 25, respectively (Kavka and Polle 2016). Similar to P. x canescens, the relative expression of several P. trichocarpa PHT1 genes was repressed by Pi (Loth-Pereda et al. 2011). Similar decreases in expression between low-P and high-P supplies were also seen for two to three PHT1 in the Solanaceae, Solanum tuberosum L. (potato) and Solanum lycopersicum (tomato) (Chen et al. 2014; Du et al. 2023). Interestingly, several cereals have low responses to Pi supply, similar to that seen for H. prostrata. These include PHT1 genes studied in the monocots Oryza sativa (rice), Zea mays (maize), Setaria italica (foxtail millet) and Lolium perenne (ryegrass) (Ceasar et al. 2014; Liu et al. 2016; Parra-Almuna et al. 2020; Wang et al. 2023). The various studies on PHT1 genes suggest that PHT1 repression from Pi supply is a widespread strategy among many eudicot species, whereas attenuated down-regulation of PHT1 expression is a less common trait exhibited by a select range of plant species such P-sensitive Australian species and may include several monocot species (Handreck 1997). Further studies are needed to determine whether attenuated PHT1 expression is a widely shared trait among plant species adapted to low nutrient soils and potentially in various monocot lineages as well as how these traits have evolved.

A second potential genetic mechanism that is also likely to contribute to P sensitivity in H. prostrata is its low capacity to down-regulate root-to-shoot translocation of Pi at elevated external supplies. HpPHT1;8 and HpPHT1;9, orthologs of AtPHT1;8 and AtPHT1;9, encode the most divergent HpPHT1 family members in H. prostrata. In A. thaliana, AtPHT1;8 and AtPHT1;9 are necessary for the root-to-shoot translocation of Pi and have been proposed to encode proteins located in endodermal and xylem pole pericycle cells of roots (Lapis-Gaza et al. 2014). The transcript abundance of these A. thaliana genes are strongly repressed by Pi in many experiments (Lapis-Gaza et al. 2014; Remy et al. 2012). Moreover, PcPHT1;9 in P. x canescens, which is orthologous to AtPHT1;8 and AtPHT1;9, was also strongly repressed by P (Kavka and Polle 2016). While transcripts from both HpPHT1;8 and HpPHT1;9 were only about 2–6% of the abundance of transcripts from the more highly expressed genes in the white roots, this low relative expression is similar to the relative expression of AtPHT1;8 and AtPHT1;9 compared with highly expressed AtPHT1 genes (Table S3) (Lapis-Gaza et al. 2014). Despite these low expression levels, AtPHT1;8 and AtPHT1;9 are indispensable for translocation of Pi to the leaves (Lapis-Gaza et al. 2014). If HpPHT1;8 and HpPHT1;9 are also involved in root-to-shoot translocation of Pi in H. prostrata, the strongly attenuated down-regulation of these transporters in white roots would likely contribute to elevated leaf P concentrations and therefore P sensitivity in H. prostrata.

Leaf [P] responses

Clearly, the levels of repression exerted by Pi on HpPHT1 genes do not prevent excess P from accumulating in leaves of H. prostrata, since the leaf P concentration of about 6.6 mg g−1 DW in the excessive-P (P12) treatment was 15- to 40-fold higher than that in the ecologically-relevant P supplies. Within A. thaliana leaves, excessive foliar P can be avoided even under high Pi supplies. The A. thaliana plants examined by Lapis-Gaza et al. (2014) had foliar P concentrations of approximately 1 mg g−1 FW for the P-depleted and 3 mg g−1 FW for the Pi resupply treatments. Neither of these supplies induced P toxicity symptoms. Therefore, A. thaliana has an approximately three-fold difference in leaf P concentration between the high- and low-Pi treatments in this study. Moreover, soil-grown P. x canescens had a maximum foliar P concentration of 6.8 mg g−1 DW, with these plants being supplied with 35 µM P per day for 63 days (Kavka and Polle 2016). This level of leaf P in P. x canescens is similar to the P12 foliar leaf P concentrations of 6.6 mg g−1 in H. prostrata. However, along with the P supply being more than twice as high as that in the current study, the soil in this study would likely accumulate Pi if nutrients were not being flushed out (Kavka and Polle 2016). Hakea prostrata would surely exhibit foliar toxicity symptoms and quite likely die if allowed to access Pi at such a rate, resulting from the low capability to control P concentrations in the mature leaves (Shane et al. 2004b). Foliar P concentrations, therefore, are clearly linked to the level of regulation of the Pi uptake systems in H. prostrata, A. thaliana and P. x canescens (Kavka and Polle 2016; Lapis-Gaza et al. 2014).

Ecological relevance of repressible HpPHT1;2 in cluster roots and white roots

An important function involved in Pi uptake by cluster roots is the transport of Pi to the rest of the plant (Shane and Lambers 2005; Srivastava et al. 2018). As the majority of Pi is generally taken up by cluster roots (Chen et al. 2023), Pi transport in cluster roots may need to be tightly regulated to prevent bursts of Pi from sink to source organs that could cause P toxicity (Lambers 2022). HpPHT1;8 and HpPHT1;9 were poorly responsive to Pi in the sampled cluster roots and white roots. Although these genes could still be regulated elsewhere in the root to regulate root-to-shoot translocation, the low Pi responsiveness makes them unlikely candidates to regulate transport of Pi taken up by cluster roots. Interestingly, HpPHT1;2 had the highest dynamic range for expression of the examined HpPHT1 genes. The expression of this gene in white roots was approximately 300-fold higher in P0.1 than in P12, even though the expression in P0.1 was relatively low, at a level similar to HpPHT1;8 and HpPHT1;9. In cluster roots, the dynamic range of HpPHT1;2 expression was only 16-fold higher in P0.1 than in P12. However, the abundance of HpPHT1;2 transcripts in cluster roots at P0.1 was among the highest for any HpPHT1. Hence, HpPHT1;2 may have an important role in Pi transport in both white roots and cluster roots. This role may be similar to the hypothesised involvement of HpPHT1;8 and HpPHT1;9 in root-to-shoot translocation of Pi. However, HpPHT1;2 may be responsible for regulation of translocation in white roots and cluster roots since HpPHT1;8 and HpPHT1;9 are not overly Pi responsive. In this case, a single P-repressible gene, HpPHT1;2, would be working in the background of two non-P-repressible genes, HpPHT1;8 and HpPHT1;9, suggesting that H. prostrata has a less effective regulation of root-to-shoot Pi translocation in cluster roots and white roots compared with regulation of Pi translocation in other plant species. These factors may contribute to the high P sensitivity in H. prostrata and may pertain to other P-sensitive Proteaceae.

Concluding remarks

Based on the present results, we conclude that the low ability of H. prostrata to down-regulate P-uptake (Shane et al. 2004c) is at least partially due to an attenuated ability to down-regulate HpPHT1 gene expression in response to higher P supply compared with the situation in many other species. Concomitantly, it appears that root growth, particularly cluster root growth, is the dominant mechanism for controlling Pi uptake in H. prostrata. This strategy means that H. prostrata is susceptible to leaf P toxicity if P supplies remain high, even after cluster root formation has been suppressed. Further study is needed to elucidate tissue locations and specific roles for the individual HpPHT1 proteins to determine the main genes and regulation patterns that contribute to P uptake and transport and potentially foliar P toxicity. Another aspect requiring further research is whether the same strategies that cause P toxicity in H. prostrata also contribute to P toxicity in other Proteaceae. Revealing the genetic and regulatory mechanisms underlying P sensitivity in Proteaceae will improve our understanding of how this highly nutrient-use-efficient plant family can thrive in nutrient-poor landscapes. This knowledge may also hold promise for reducing global fertiliser use, allowing the transfer of similar traits into crop species.