Contrasting nickel and zinc hyperaccumulation in subspecies of Dichapetalum gelonioides from Southeast Asia

Hyperaccumulator plants have the unique ability to concentrate specific elements in their shoot in concentrations that can be thousands of times greater than in normal plants. Whereas all known zinc hyperaccumulator plants are facultative hyperaccumulators with only populations on metalliferous soils hyperaccumulating zinc (except for Arabidopsis halleri and Noccaea species that hyperaccumulate zinc irrespective of the substrate), the present study discovered that Dichapetalum gelonioides is the only (zinc) hyperaccumulator known to occur exclusively on ‘normal’ soils, while hyperaccumulating zinc. We recorded remarkable foliar zinc concentrations (10 730 µg g−1, dry weight) in Dichapetalum gelonioides subsp. sumatranum growing on ‘normal’ soils with total soil zinc concentrations of only 20 µg g−1. The discovery of zinc hyperaccumulation in this tropical woody plant, especially the extreme zinc concentrations in phloem and phloem-fed tissues (reaching up to 8465 µg g−1), has possible implications for advancing zinc biofortification in Southeast Asia. Furthermore, we report exceptionally high foliar nickel concentrations in D. subsp. tuberculatum (30 260 µg g−1) and >10 wt% nickel in the ash, which can be exploited for agromining. The unusual nickel and zinc accumulation behaviour suggest that Dichapetalum-species may be an attractive model to study hyperaccumulation and hypertolerance of these elements in tropical hyperaccumulator plants.

• Dichapetalum gelonioides subsp. tuberculatum is a strong Ni hyperaccumulator with up to 31 700 µg g −1 Ni • Dichapetalum gelonioides subsp. tuberculatum can also attain high Zn concentrations of up to 3990 µg g −1 • Dichapetalum gelonioides subsp. pilosum is a strong Zn hyperaccumulator with up to 10 600 µg g −1 Zn • Dichapetalum gelonioides subsp. sumatranum is also a strong Zn hyperaccumulator with up to 10 500 µg g −1 Zn • Dichapetalum grandifolium has elevated Zn concentrations of up to 1300 µg g −1 Herbarium specimens were also measured with handheld XRF at the Queensland Herbarium for several Asia-Pacific species of Dichapetalum, including D. papuanum, D. timoriense, D. tricapsulare and D. sessiliflorum, none of which hyperaccumulated Ni or Zn. However, a specimen of D. vitiense from Fiji (collected in 1947 at Vanua Levu, on the summit ridge of Mt Numbuiloa, east of Lambasa, 500-590 m asl in forest) was confirmed as a Zn hyperaccumulator with 7800 μg g −1 foliar Zn. This finding raises the prospect that Dichapetalum species other than D. gelonioides could be Zn hyperaccumulators.
Dichapetalum gelonioides is a tropical species that occurs throughout Southeast Asia and attains high biomass, thereby overcoming the main drawbacks that limit the use of A. halleri and N. caerulescens for biofortification in tropical areas. Apart from the biofortification potential, D. gelonioides may be useful to advance our knowledge on hyperaccumulation of trace elements, particularly Zn. All Zn hyperaccumulator plant species known to date are facultative hyperaccumulators, and hence the majority occur primarily on normal soils (where they do not hyperaccumulate Zn), and have some populations on metalliferous soils where they do hyperaccumulate Zn 41 . Therefore, these Zn hyperaccumulators behave essentially as 'Indicators' (sensu Baker 49 ) in that the level of accumulation in the shoot is strongly dependant on soil Zn concentrations. The main exceptions are A. halleri and N. caerulescens that have populations on normal soils that are also able to hyperaccumulate Zn 41 . Dichapetalum gelonioides is the only (Zn) hyperaccumulator known to occur exclusively on normal soils, while hyperaccumulating Zn. As such, Dichapetalum gelonioides provides an attractive model to study Zn tolerance, uptake, translocation, accumulation and detoxification, and may ultimately be useful for applications in the field of biofortification. This study aims to decipher the extent and variability of Ni and Zn hyperaccumulation traits in wild D. gelonioides subspecies occurring in Sabah.

Results
Elemental concentrations in different parts of Dichapetalum gelonioides plants. The  have low Ni concentrations (mean is 10 and 15 µg g −1 respectively) (Fig. 1). On the contrary, the Zn concentrations are substantially elevated in the leaves of D. subsp. pilosum and D. subsp. sumatranum (>3000 µg g −1 ), but that of the D. subsp. tuberculatum are below 300 µg g −1 (Fig. 1). The concentrations of the major elements, Ca, K, Mg and S are relatively elevated (mean concentrations >1000 µg g −1 ), whereas the concentrations of the trace elements Co and Fe are uniformly low (mean < 50 µg g −1 ) in the leaves of all subspecies. Whereas the concentrations of Mn are elevated in the leaves of D. subsp. pilosum (>500 µg g −1 ), that of D. subsp. tuberculatum are low (mean <50 µg g −1 ). The highest Ni and Zn concentrations are recorded in the old leaves of D. subsp. tuberculatum (23 480 µg g −1 ) and D. subsp. sumatranum (9780 µg g −1 ) respectively, followed by young leaves with up to 18 230 µg g −1 for D. subsp. tuberculatum and 7720 µg g −1 for D. subsp. sumatranum respectively. The Ni concentrations in the branches and twigs of D. subsp. tuberculatum are substantially enriched (>1000 µg g −1 ). For D. subsp. pilosum and D. subsp. sumatranum, the Zn concentrations in the branches and twigs are elevated, but the values are below 3000 µg g −1 .
Unripe and matured fruits of D. subsp. sumatranum are also high in Zn (1955 and 2785 µg g −1 respectively). The bark/phloem of D. subsp. sumatranum has exceptionally high Zn concentrations (8465 µg g −1 ), whereas that of D. subsp. pilosum is elevated in Zn but below 3000 µg g −1 . Whilst the roots of D. subsp. tuberculatum   Light microscopy of woody stems of Dichapetalum gelonioides. Light microscopy images (Fig. 2) show a whitish phloem in the woody stem of D. subsp. pilosum whilst in D. subsp. tuberculatum the phloem is greenish, mainly from the high concentrations of Zn and Ni ions respectively. The images show that Ca crystals are more abundant in the medullary rays of the stem of D. subsp. pilosum than in D. subsp. tuberculatum (Fig. 2). Synchrotron X-ray fluorescence microscopy investigation. Figure  In the growth bud of D. subsp. pilosum (Fig. 4), Zn is strongly enriched in the node and the emerging leaves where it reaches up to 0.20 wt%, whereas there is a minor enrichment in the stem with up to 0.08 wt%. Interestingly, the distribution of Ni mirrors that of Zn, but occurs at much low concentrations (<300 µg g −1 ). Potassium is strongly enriched in the stem and the young leaves with minor enrichment in the node, whereas Ca is strongly enriched in the node but has relatively low concentration in the stem and in the emerging leaves.
Zinc is distributed throughout the leaf of D. subsp. pilosum with concentrations reaching up to 0.30 wt%, whereas it is distributed at lower concentration in D. subsp. tuberculatum with concentrations below 100 µg g −1 (Fig. 5). Nickel is particularly low in the phloem of the veins of D. subsp. pilosum. In D. subsp. tuberculatum Ni is distributed at low concentrations throughout the leaf, but strongly enriched in the phloem of the mid-vein where it exceeds 2.00 wt%. Potassium is strongly enriched in the phloem of the mid-vein of both D. subsp. pilosum and D. subsp. tuberculatum. Manganese is distributed more or less uniformly in the leaf of both D. subsp. pilosum and D. subsp. tuberculatum at very low concentrations. The distribution of Ca mirrors that of K, albeit with Ca-oxakate deposits lining the veins.
Rhizosphere soil chemistry of Dichapetalum gelonioides. The rhizosphere soil of D. gelonioides subsp.
tuberculatum is near neutral (pH 6.01) as expected for well-buffered ultramafic soils rich in Mg, whereas that of the D. subsp. pilosum and D. subsp. sumatranum is acidic (pH 5.05 and 4.64, respectively), typical for leached tropical soils ( Table 3). The total Ni concentration is high in the ultramafic soil on which D. subsp. tuberculatum occurs, unlike that of D. subsp. pilosum and D. subsp. sumatranum growing on sandstone soils, where low total soil Ni concentrations (8.     subsp. tuberculatum contained as high as 11.3 wt% Ni, which is significantly higher than in traditional lateritic ore (which typically has <1 wt% Ni). Similar high-grade ash compositions are recorded in the two most promising species of tropical Ni agromining, Phyllanthus rufuschaneyi (12.7 wt%) and Rinorea cf. bengalensis (5.50 wt% Ni) 51 . The unique 'bio-ore' composition, coupled with the intrinsically high Ni grade, make it possible to extract even higher value products such as Ni catalysts for the organic chemistry industry 52 and electrochemical Ni products 53 , although smelting of Ni metal is in itself technically feasible 6 . We must add that despite the high purity of the 'bio-ore' relative to conventional ore, significant concentrations of Ca (9.49-10.1 wt%), Mg (2.07-2.48 wt%), K Similarly high values (53 900 µg g −1 ) have been reported in A. halleri from Germany growing on non-metalliferous soils 44 . Zinc hyperaccumulation occuring on soils with 'normal' Zn concentrations is known only from these two European species (A. halleri and N. caerulescens). Reeves et al. 41 analysed field collected specimens of N. caerulescens and the corresponding 'normal' soils in France and Luxembourg, and found remarkable foliar Zn concentrations of 3230-8890 µg g −1 occurring on soils with Zn concentrations of only 115-274 µg g −1 . Bert et al. 29 also reported 10 876 ± 578 µg g −1 Zn concentrations in the aerial part of wild A. halleri plants in Germany occurring on soils with 201 ± 30 µg g −1 Zn concentrations. The total Zn concentrations in 'normal' soils in Sabah are typically between 1.20 to 150 µg g −1 55 . The present study reveals a remarkable foliar Zn concentration of 10 730 µg g −1 in D. subsp. sumatranum growing on 'normal' soils with total Zn concentration of only 20 µg g −1 . It is noteworthy that the concentrations we report are not the highest for this species, but that Baker et al. 46 found as high as 14 000 µg g −1 and 25 000 µg g −1 in herbarium specimen of D. subsp. sumatranum and D. subsp. pilosum respectively (Fig. 6), showing that Dichapetalum species are among the strongest Zn hyperaccumulators globally. Regarding the (Zn) Bioaccumulation Coefficients (BC) reported for Zn hyperaccumulator species, the values for species occurring on metalliferous soils are relatively lower than that of the non-metalliferous soils 41,44 . Previously, the highest BC reported for a Zn hyperaccumulator species was recorded in A. halleri occurring on non-metalliferous soils with a BC of ~1270 44 . The present study also reveals an exceptional BC of ~500 for D. subsp. sumatranum, further confirming the extraordinary accumulation behaviour of Dichapetalum species. We discovered that Dichapetalum gelonioides is the only (Zn) hyperaccumulator known to occur exclusively on normal soils, while hyperaccumulating Zn.
In Dichapetalum, Zn accumulation at extreme concentrations is not only limited to the leaves, but in other plant tissues as well, even though the distribution is uneven. The present study reveals as high as 8465 µg g −1 Zn concentrations in the phloem/bark of D. subsp. sumatranum. This exceptionally high Zn concentration in the phloem may be transported to the fruits and seeds. The mature fruit of D. subsp. sumatranum has high Zn concentrations of up to 2785 µg g −1 . The high Zn concentration in the phloem and the fruits of D. subsp. sumatranum could have implications for Zn biofortification of edible crop plants. For crop plants, Zn concentrations decrease in the order of roots >shoots >fruits and seeds 56 . Notably, low Zn mobility in phloem limits the Zn concentrations in fruits and seeds, and these phloem-fed tissues rarely achieve Zn concentrations greater than 30-100 µg g −1 15 . As a result, Zn concentrations are very low in grain, seed, fruit or tuber crops [57][58][59] . Consequently, Zn deficiency disorders are more prevalent in population with diets dominated by such crops 60 . Zinc is an essential trace element for all organisms, and particularly important for human nutrition 57,58,61 . The Institute of Medicine USA 62 recommends a daily intake of 8-13 mg Zn, but up to 1/3 of the global population has inadequate dietary Zn intake because of limited access to Zn-rich foods, and consequently, Zn deficiency is a major health problem in developing countries [63][64][65][66] . Particularly, it is more problematic in Southeast Asia because of the consumption of diets low in Zn, mainly polished rice. Limited dietary Zn intake can be minimised by improving the available Zn in the diet, and biofortification of edible crops is a sustainable strategy especially in developing countries 15,[67][68][69] . The target Zn concentrations are 28 µg g −1 in rice, and 38 µg g −1 in wheat grain and maize 68 . Zinc hyperaccumulators have the potential to advance biofortification, because insights into the biomolecular mechanisms of Zn acquisition may be applied to food crops 15 , and hyperaccumulator plant biomass may be used as a   dietary supplement 14 . However, A. halleri and N. caerulescens are both temperate region plants with low biomass, and hence unsuited for cultivation in tropical regions where human Zn deficiency is prevalent. Hence, the discovery of a Zn hyperaccumulator plant species on 'normal' soils in Sabah provides a strong incentive to advance Zn biofortification measures in Southeast Asia. Insights gained from the mechanisms underlying the remarkable Zn translocation and accumulation especially in the phloem and phloem-fed tissues may be particulary useful. Evidence from transcriptome data suggests that Zn hyperaccumulator plant species have an enhanced capacity to protect cells from metal toxicity 22,27 . On the other hand, crop plants maintain low Zn concentration in the phloem sap as a strategy to avoid cellular toxicity 58 . Therefore, identifying the genes involved in the Zn hyperaccumulation in D. subsp. sumatranum could enable the breeding of genetically modified edible crops with enhanced abilities, especially translocation of Zn in the phloem. The increased Zn tolerance and mobility in phloem of these improved cultivars would therefore enrich the phloem-fed tissues (such as seeds/grains), and ultimately increase dietary Zn intake in Southeast Asia. Another approach to increase dietary Zn intake in Southeast Asia may be to introduce the biomass of hyperaccumulators, such as D. subsp. sumatranum and D. subsp. pilosum, as a food supplement 14 . Considering a recommended intake of 8 mg Zn (Institute of Medicine USA 62 ), an average 33% efficiency of human uptake 70 , and a leaf Zn concentration of 9000 µg g −1 in the leaves of D. subsp. sumatranum (Table 2), consuming 0.5 g dry material would meet one fifth of the daily requirements. As Zn in the leaves of Zn hyperaccumulators is likely bioavailable, human Zn uptake from these materials should not be an issue. However, there are major food safety concerns regarding the use of Zn hyperaccumulator plant species as food because A. halleri and N. caerulescens can also accumulate Cd in their leaves 41 . Cadmium is highly toxic to humans, and obviously consuming plant tissues with high bioavailable Cd is not recommended for human health 71 . Therefore, there is a need to test whether Dichapetalum-species occurring on 'normal' soils will accumulate Cd. Furthermore, many Dichapetalum-species are extremely toxic to livestock because they contain fluorinated compounds such as fluoroacetic acid and ω-fluorinated fatty acids [72][73][74][75][76] . Therefore, these fluorinated compounds will need to be detoxified before they are introduced as food supplements for human consumption, and toxicological assays are clearly required. Some Dichapetalum-species can also contain fluorine-free compounds, such as the triterpenoids group of dichapetalins that are being intensively investigated because of their strong cytotoxic activity, and potential application in anti-cancer drugs [77][78][79][80][81] .
Apart from the potential for agromining and biofortification applications, Dichapetalum-species can advance our understanding on both Ni and Zn hyperaccumulation. The mechanisms involved in the hyperaccumulation of trace elements are not fully understood, but the key processes include uptake by roots and xylem loading as well as sequestration in the leaf cells. Whether hyperaccumulation is primarily driven by root processes, or hypertolerance is determined by shoot processes remains inconclusive, at least in tropical hyperaccumulator plant species. Grafting experiments in which the rootstocks and shoot scions are reciprocally grafted from D. subsp. pilosum or D. subsp. sumatranum (Zn hyperaccumulating subsp.) and D. subsp. tuberculatum (Ni hyperaccumulating subsp.) may unravel the distinct roles of root and shoot processes in hyperaccumulation and hypertolerance. In addition, reciprocal dosing of Dichapetalum-species in which the Zn hyperaccumulating subsp. is dosed with Ni, and vice versa may provide more insights into the uptake, translocation, tolerance and accumulation of these trace elements. Furthermore, Ni and Zn stable isotope tracing experiments can provide useful information on both uptake and transfer mechanisms within the plant-soil systems. Therefore, Dichapetalum-species are an attractive model to study both Ni and Zn hyperaccumulation and hypertolerance in tropical hyperaccumulator plant species.

Materials and Methods
Herbarium XRF scanning. The Niton XL3t 980 analyser (Thermo-Fisher Scientific) uses a miniaturised X-ray tube (Ag anode; 6-50 kV, 0-200 µA max) as its excitation source. The X-ray tubes irradiates the sample with a stable source of high-energy X-rays, and fluorescent X-rays are detected, identified and quantified by the inbuilt Silicon Drift Detector (SDD) with ~185 eV, up to 60 000 cps, 4 µs shaping time. The instrument uses Compton normalisation for quantification, appropriate for the relatively low elemental concentrations found in plant material. A total of 590 dried plant samples originating from Sabah, Malaysia 82 were used for the calibration. From each sample, a 6-mm diameter leaf disc (to match the XRF beam width) was extracted using a paper punch. A square of ~99.7% pure titanium (2 mm thick × 10 cm × 10 cm; Sigma-Aldrich 369489-90G) was used behind the specimens to provide a uniform background and block penetrating X-rays. XRF measurements were carried out in the 'Soils Mode' for 60 s. After scanning, the leaf samples were weighed, digested and analysed with Inductively Coupled Plasma Atomic Emission Spectroscopy (ICP-AES), following the procedures described below. Correction factors were derived by linear regression of XRF data against corresponding ICP-AES measurements.
In total, 91 herbarium species were measured at the herbarium of the Forest Research Centre in Sepilok, Sabah, Malaysia. This comprised two species: Dichapetalum grandifolium and Dichapetalum gelonioides (subsp. tuberculatum, pilosum, sumatranum). Each specimen was measured for 30 seconds in 'Soil Mode' and the raw XRF data was corrected using the regression formulas obtained from the calibration.  (Fig. 7). These samples were dried at 70 °C for five days in a drying oven and subsequently packed for transport to Australia and gamma irradiated at Steritech Pty. Ltd. in Brisbane following Australian Quarantine Regulations. The dried plant tissue samples were subsequently ground and ~300 mg material digested using 4 mL HNO 3 (70%) and 1 mL H 2 O 2 (30%) in a microwave oven (Milestone Start D) for a 45-minute programme and diluted to 30 mL with ultrapure water (Millipore 18.2 MΩ·cm at 25 °C). Further, bulk samples of D. gelonioides subsp. tuberculatum (leaves), D. subsp. sumatranum (leaves), D. subsp. sumatranum (stems), and D. subsp. pilosum (twigs and leaves) were collected. After oven-drying at 70 °C, subsamples of ~750 mg were ashed in a muffle furnace at 550 °C for 5 hrs and cooled to room temperature. Samples were weighed before and after ashing for gravimetric analysis. The ash was subsequently dissolved in 37% HCl (10 mL per sample), and analysed with ICP-AES as described below.
Rhizosphere soil samples were collected from near the roots of the Dichapetalum plants. Soil sub-samples (~300 mg) were digested using 9 mL 70% HNO 3 and 3 mL 37% HCl per sample in a digestion microwave (Milestone Start D) for a program of 1.5 hours, and diluted to 45 mL with ultrapure water before analysis to obtain pseudo-total elemental concentrations. Soil pH was obtained in a 1 to 2.5 soil to water mixture after 2 hr shaking. Exchangeable trace elements were extracted in 0.1MSr(NO 3 ) 2 at a soil:solution ratio of 1:4 (10 gram soil with 40 mL solution) and 2 hr shaking time (adapted from Kukier & Chaney 83 ). As a means of estimating potentially phytoavailable trace elements, the DTPA-extractant was used according to Becquer et al. 84 , which was adapted from the original method by Lindsay and Norvell 85 , with the following modifications: excluding triethanolamine (TEA), adjusted at pH 5.3, 5 g soil with 25 mL extractant, and extraction time of one hr. The plant digests and soil digests/ extracts were analysed with ICP-AES (Varian Vista Pro II) for Ni, Co, Cr, Cu, Zn, Mn, Fe, Mg, Ca, Na, K, S and P; the Certified Reference Materials (CRMs) are given in Supplementary

Collection and preparation of samples for Synchrotron X-ray Fluorescence Microscopy (XFM).
Plant tissue samples (roots, wood, leaves and phloem tissue) and rhizosphere soil samples were collected in the native rainforest habitat in Malaysian Borneo (Sepilok-Kabili Forest Reserve for D. subsp. sumatranum and D. subsp. pilosum and Mount Silam Forest Reserve for D. subsp. tuberculatum, in Sabah, Malaysia). Tissue samples intended for synchrotron analysis were excised with a razor blade, and immediately shock-frozen using a metal mirror technique in which the samples were pressed between a block of Cu-metal cooled by liquid N 2 and a second Cu-metal block attached to a Teflon holder. This ensured extremely fast freezing of the plant tissue samples to prevent cellular damage by ice crystal formation. The samples were then wrapped in Al-foil, and transported in a cryogenic container (kept at <180 °C).
The samples required freeze-drying because they could not be kept frozen during the X-ray fluorescence microscopy (XFM) analysis. The samples were freeze-dried (using a Thermoline freeze-dryer) employing a long process of 4 days in which the temperature was stepwise increased, starting at −196 °C on a liquid N 2 pre-cooled block (with the freeze-dryer set to −85 °C) until room temperature, to limit morphological changes of the tissues and elemental re-distribution. Synchrotron X-ray Fluorescence Microscopy. The X-ray fluorescence microscopy (XFM) beamline employs an in-vacuum undulator to produce a brilliant X-ray beam. An Si(111) monochromator and a pair of Kirkpatrick-Baez mirrors delivers a monochromatic focused incident beam onto the specimen 86 . The Maia detector uses a large detector array to maximize detected signal and count-rates for efficient imaging. Maia enables high overall count-rates and uses an annular detector geometry, where the beam passes though the detector and strikes the sample at normal incidence 87,88 . This enables a large solid-angle (1.2 steradian) to be achieved in order to either maximize detected signal or to reduce the dose and potential damage to a specimen 89 . Maia is designed for event-mode data acquisition, where each detected X-ray event is recorded, tagged by detector number in the array, position in the scan and other metadata 90 . This approach eliminates readout delays and enables arbitrarily short pixel times (typically down to 0.1 ms) and large pixel count to be achieved for high definition imaging (typically 10-100 M pixels). The freeze-dried samples were mapped by sampling at intervals ranging from 2-10 μm and 0.5-5 ms transit time. The XRF event stream was analysed using the Dynamic Analysis method 91,92 as implemented in GeoPIXE 93 .