Difference between revisions of "Journal:A spectroscopic study to assess heavy metals absorption by a combined hemp-spirulina system from contaminated soil"

From CannaQAWiki
Jump to navigationJump to search
(Saving and adding more.)
(Saving and adding more.)
Line 121: Line 121:


==Results and discussion==
==Results and discussion==
NMR analysis (1D <sup>1</sup>H NOESY experiments) allowed for the identification of a pool of metabolites contained in typical aqueous extracts (pH = 4.2) of leaves (Fig. 1a) and stems (Fig. 1b) collected from a plant of hemp cultivated under controlled conditions in uncontaminated soil (see ESI for the full list of metabolites, Table S3). The main representative classes of metabolites included organic acids (i.e., lactic, citric, tartaric, fumaric, formic, and acetic acids), amino acids (i.e., alanine, asparagine, γ-aminobutyric acid, arginine, glutamine, isoleucine, proline, tyrosine, leucine), and carbohydrates (i.e., β-glucose, α-glucose, fructose, α-galactose, β-galactose). Moreover, 1D <sup>1</sup>H NOESY spectra also contained signals related to trigonelline, choline, and [[ethanol]].


[[File:Fig1 Musio EnviroAdv2022 7.jpg|1200px]]
{{clear}}
{|
| STYLE="vertical-align:top;"|
{| border="0" cellpadding="5" cellspacing="0" width="1200px"
|-
  | style="background-color:white; padding-left:10px; padding-right:10px;"| <blockquote>'''Fig. 1''' Typical 1D <sup>1</sup>H NOESY spectra (Bruker Avance 400 MHz, D<sub>2</sub>O) of aqueous extracts of leaves ('''a.''') and stems ('''b.''') of ''Cannabis sativa'' L. The spectral region containing the residual water signal (4.78 ppm) was hidden. The spectral regions included in rectangles were expanded for clarity.</blockquote>
|-
|}
|}
A holistic approach was applied during the present study by combining the information provided by the 1D <sup>1</sup>H NOESY experiments, the ICP-AES quantitative analysis of residual metal, and the data derived from the morphometric inspection.
As a first task, the relation between the variations in the metabolic composition and the calculated amounts of residual heavy metals in the plant was considered. Thus, the OPLS analysis was performed considering the spectral buckets as ''x''-variables, and the metal content and plant biomass as ''y''-variables (see ESI for OPLS model parameters, Table S4). Such a model was characterized by five predictive components P1-P5 that explained 59.5% of ''x''-variance (R<sup>2</sup>X[cum] = 0.595), 92.2% of ''y''-variance (R<sup>2</sup>Y[cum] = 0.922) and 87.6% of ''y''-variance modelled by ''x''-variables (R<sup>2</sup> = 0.876), and eight orthogonal components O1-O8 that explained about 38% of the ''x''-variance unrelated to ''y''-variance (R<sup>2</sup>X[cum] = 0.383). Two main separate clusters of observations were noticeable along P1. They corresponded to the two parts of the plant, namely the leaves (LE) and the stems (ST) (Fig. 2a, LE vs ST). Inside each group, two evident subgroups were distinguishable along P2, which were identified as the vigorous plants (V) and the weak ones (W) (Fig. 2b, V vs W). The components P1 and P2 explained 41 and 9.8% of the information in ''x''-space (R<sup>2</sup>X[P1] = 0.41 and R<sup>2</sup>X[P2] = 0.098), suggesting an important variability of the metabolic composition along with the different tissues of the plant and according to the wellness of the plant. Indeed, the same components P1 and P2 explained 48.5 and 15.2% of the information in ''y''-space (R<sup>2</sup>Y[P1] = 0.485 and R<sup>2</sup>Y[P2] = 0.152), suggesting a differential distribution of the ''y''-variables, namely the metals concentration and the plant biomass, according to the different tissue of the plant and the health status of the plant (Fig. 2c). The visual inspection of the original harvested plants (see ESI for further details, Fig. S1), combined with the plant biomass values, suggested that the samples included in the subgroup with t[2]<0 (Fig. 2b, red scores) corresponded to the less healthy plants (weak, W), whereas the observations included in the subgroup with t[2]>0 corresponded to the more vigorous plants (V, Fig. 2b, yellow scores). The analysis of the loading plot (pq[1] vs pq[2]) allowed the identification of the most significant variables towards the distribution of the scores, including the metal concentration, biomass amount, and spectral regions (Fig. 2c).
[[File:Fig2 Musio EnviroAdv2022 7.jpg|1200px]]
{{clear}}
{|
| STYLE="vertical-align:top;"|
{| border="0" cellpadding="5" cellspacing="0" width="1200px"
|-
  | style="background-color:white; padding-left:10px; padding-right:10px;"| <blockquote>'''Fig. 2''' OPLS-DA applied to spectral data (''x''-variables) and metal content and plant biomass data (''y''-variables). ('''a.''') Scores plot displaying t[1] vs t[2], the observations are colored according to the tissues of the plants: leaves (green circle, LE), stems (blue square, ST). ('''b.''') Scores plot t[1] vs t[2], the observations are colored according to the health status of the plants: vigorous (yellow circle, V), weak (red square, W). ('''c.''') Loading plot displaying the relationship between the ''x''-variables (spectral regions) and the ''y''-variables (metal content and plant biomass) for the first and second predictive components pq[1] vs pq[2] (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.).</blockquote>
|-
|}
|}





Revision as of 19:33, 11 September 2022

Full article title A spectroscopic study to assess heavy metals absorption by a combined hemp-spirulina system from contaminated soil
Journal Environmental Advances
Author(s) Musio, Biagia; Ahmed, Elhussein M.F.M.H.; Antonicelli, Marica; Chiapperini, Danila; Dursi, Onorfrio; Grieco, Flavia; Latronico, Mario; Mastrorilli, Piero; Ragone, Rosa; Settanni, Raffaele; Triggiani, Maurizio; Gallo, Vito
Author affiliation(s) Polytechnic University of Bari, Innovative Solutions S.r.l., International Centre for Advanced Mediterranean Agronomic Studies of Bari, ApuliaKundi S.r.l.
Primary contact Email: vito dot gallo at poliba dot it
Year published 2022
Volume and issue 7
Article # 100144
DOI 10.1016/j.envadv.2021.100144
ISSN 2666-7657
Distribution license Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International
Website https://www.sciencedirect.com/science/article/pii/S2666765721001150
Download https://www.sciencedirect.com/science/article/pii/S2666765721001150/pdfft (PDF)

Abstract

The efficiency of hemp (Cannabis sativa L.) in remediating sites contaminated with heavy metals has received great attention in recent years. The main advantage of this technology relies on its inherent sustainability, with a potential re-utilization of the significant amount of produced biomass, which acts as a valuable flow resource. In this study, a combined system consisting of Cannabis sativa L. (hemp) and the blue-green alga Arthrospira platensis (spirulina) was tested to clean up soils contaminated with cadmium, chromium, copper, nickel, lead, and zinc. The application of non-targeted nuclear magnetic resonance spectroscopy (NMR) methods combined with inductively coupled plasma atomic emission spectroscopy (ICP-AES) quantification provided an efficient strategy for detecting residual heavy metals within plant tissues and soil. Importantly, non-targeted metabolomic analysis helped to reveal the relationships between metabolites distribution in hemp tissues and the sequestered metals. It was demonstrated that hemp accumulates copper, chromium, nickel, and zinc preferentially in the leaves, while lead is distributed mainly in the stems of the plant. Moreover, it was found that, at higher concentrations, spirulina acts as a growth promoter, contributing to an increase in the final generated biomass. Results reported in this work indicate that the hemp-spirulina system represents a suitable tool for remediation of metal contaminated soils by modulating biomass production and metals uptake.

Keywords: non-targeted nuclear magnetic resonance, phytoremediation, phycoremediation, Arthrospira platensis, Cannabis sativa L., metal quantification

Graphical abstract:

GA Musio EnviroAdv2022 7.jpg

Introduction

Dispersion of heavy metals in soils is an age-old problem deriving from both natural and anthropic sources. (Awa and Hadibarata, 2020) Among the anthropic contribution to soil contamination by metals, land application of treated wastewater, sewage sludge, fertilizers, and industrial activities are major concerns. (Vareda et al., 2019) Unbalanced amounts of heavy metals may cause perturbation of soil parameters with consequent toxic effects on plants, in the nearby water supplies, and, ultimately, in the whole food chain. (Arora et al., 2008; Kumar et al., 2019; Manzoor et al., 2018) Typically, elements such as copper (Cu), nickel (Ni), zinc (Zn), and chromium (Cr) are biologically essential for plant growth but become toxic for animals and plants when their concentrations exceed certain threshold levels. (Edelstein and Ben-Hur, 2018; Rizvi et al., 2020; Tiwari and Lata, 2018) Other heavy metals often found in contaminated soils, such as cadmium (Cd) and lead (Pb) are not essential for plants growth, and many studies have associated their presence with neurological and endocrinological toxicity for humans, along with carcinogenic effects. (Ali and Khan, 2019; Pratush et al., 2018; Rehman et al., 2018)

Since heavy metals are not biodegradable, they tend to accumulate in the environment, becoming a high risk for biota over several years after their introduction in an ecosystem. (Olsson et al., 1998; Tchounwou et al., 2012; Zwolak et al., 2019) The search for new solutions that can remediate soil contaminated by heavy metals is a critical prerequisite for the sustainable development of agriculture (Edelstein and Ben-Hur, 2018; ; Vardhan et al., 2019; Wuana and Okieimen, 2011), thus representing a topic of paramount importance. The most consolidated strategies to remediate such contaminated soils include physical and chemical approaches like isolation, through capping and subsurface barriers; immobilization, by solidification/stabilization, vitrification, and chemical treatment; physical separation; and extraction, by soil washing, pyrometallurgical extraction, in situ soil flushing, and electrokinetic treatment. (Dhaliwal et al., 2020; Gong et al., 2018; Gusiatin et al., 2020; Qin et al., 2020) However, alternative approaches are gaining greater attention as they combine cost-effectiveness, sustainability, low toxicity, and mobility decrease. They include bioaccumulation, phytoremediation (e.g., phytoextraction, phytostabilization, and rhizofiltration), bioleaching, and other biochemical processes in which living organisms such as plants or microbes are used to clean contaminants from an area.

In particular, phytoremediation is attracting the attention of the scientific community, since it has been demonstrated to be a cost-effective solution for the remediation of contaminated sites, and, in the meantime, to be a feasible method for bio-fixation of CO2, resulting in highly sustainable technology. (Awa and Hadibarata, 2020) The ability to absorb heavy metals generally depends on the biomass produced, as well as on the ability of the plant to accumulate and translocate heavy metals in its biomass. (Eid and Shaltout, 2016; Hernández-Allica et al., 2008; Pachura et al., 2016) According to recent scientific literature, a good candidate for phytoremediation of soil contaminated by heavy metals is the hemp plant. (Ahmad et al., 2016; Morin-Crini et al., 2019; Zielonka et al., 2020) Kompolti, also known as hemp, the non-psychoactive variety of Cannabis sativa L., is an annual dioecious high-yielding industrial crop, and it is mainly grown for its fibers and seeds, generally being used for textiles, clothing, insulation, biodegradable plastics, food, animal feed, and biofuel production. (Adesina et al., 2020; Crini et al., 2020; Schluttenhofer and Yuan, 2017; Vasantha Rupasinghe et al., 2020) Hemp possesses some characteristics that make it quite suitable for phytoremediation, such as high biomass, long roots, and a favorably short industrial life cycle of 180 days. Importantly, hemp demonstrates a strong capability to sequester heavy metals like cadmium, zinc, lead, nickel, copper, and chromium when they are present in contaminated soil and water. (Citterio et al., 2003; Galić et al., 2019; Piotrowska-Cyplik and Czarnecki, 2003; Zielonka et al., 2020)

Another attractive approach for the remediation of contaminated sites is the application of bioleaching technology, which uses direct metabolism or by-products of microbial processes to uptake heavy metals adsorbed onto the soil surface and to transform them so that the elements can be extracted when water is filtered through. Bioleaching has several advantages over conventional physical and chemical strategies, such as low cost, environmental sustainability, few hazardous characteristics of waste/sludge, low energy demand, and absence of toxic chemicals. (Bosecker, 1997; Drobíková et al., 2015; Mishra et al., 2005; Okoh et al., 2018; Rawlings, 2002; Sun et al., 2021)

Additionally, phycoremediation, which involves eukariotic algae and cyanobacteria in remediation processes, has been extensively applied to the treatment of wastewater. (Awa and Hadibarata, 2020) However, its application to the remediation of sediments and soils contaminated by heavy metals is less documented. Among the cyanobacteria, Arthrospira platensis possesses excellent chelating properties both towards heavy metals present in humans and towards those present in soil, water, and sludge. (Balaji et al., 2014; Bhattacharya, 2020; Konig-péter et al., 2015; Nalimova et al., 2005; Zinicovscaia et al., 2019, 2016) The dried biomass of Arthrospira platensis is commonly known as "spirulina," and it finds many applications in agriculture as a plant growth promoter, enhancing growth, increasing yield, and speeding up seed germination. (Tripathi et al., 2008; Wuang et al., 2016) Recently, the employment of this blue-green alga to uptake heavy metals in contaminated sites has been explored. (Cepoi et al., 2020; Wuang et al., 2016) The presence of a chloroplast-type ferredoxin in the active center has been reported as responsible for the chelating capability of spirulina (Tsukihara et al., 1978), whereby its efficiency is affected by many physical and chemical factors such as initial metal concentration, dosage, adsorption time, temperature, and pH. (Şeker et al., 2008)

The present study aims at both exploring the ability of the unreported combined use of hemp and spirulina to uptake six selected heavy metals (Cd, Ni, Cr, Pb, Cu, Zn) from artificially contaminated soil and investigating, under controlled plant growing conditions, their distribution into the plant tissues. Specifically, hemp was chosen as the main agent for biological remediation, and spirulina was added as an enhancer of both the plant growth and the translocation of heavy metals in the hemp. The application of a non-targeted nuclear magnetic resonance spectroscopy (NMR) approach combined with an estimation of the residual metals by inductively coupled plasma atomic emission spectroscopy (ICP-AES) into the cultivation soil and within the different tissues of the plant was applied in view of gathering useful information on the efficiency of the integrated hemp-spirulina system. Obtaining this information is crucial for the potential re-utilization of the hemp plant or shoots of it, after the phytoremediation stage, for alternative usages, like production of bio-materials for the textile, construction, and bio-fuel industries.

Materials and methods

Materials

3-(Trimethylsilyl)-2,2,3,3-tetradeutero-propionic acid sodium salt (TSP, CAS N. 24493-21-8, 99 %D, Armar Chemicals, Döttingen, Switzerland), hydrochloric acid (HCl, 37%, CAS N. 7647-01-0; ≥ 99.5%, Sigma-Aldrich, Milan, Italy), sodium oxalate (Na2C2O4, CAS N. 62-76-0; ≥99.5%, Sigma-Aldrich, Milan, Italy), sodium azide (NaN3, CAS N. 26628-22-8; ≥99.5%, Sigma-Aldrich, Milan, Italy), and deuterium oxide (D2O, CAS. N. 7789-20-0, 99.86 %D, Eurisotop, Saclay, France) were used for sample preparation. NMR tubes (Norell 509-UP 7) were provided by Norell, Landisville NJ, United States. The NMR samples were prepared using an automated system for liquid handling (SamplePro Tube, Bruker BioSpin).

The soil used during cultivation was Plagron Lightmix (Plagron, Ospel, The Netherlands), which had pH 6-7, electric conductivity (E.C. 1:5) 310-470 µS•cm−1 and (E.C. 1:1.5) 0.7-1.1 µS•cm−1, NPK (12-14-24) 1.5 kg•m−3, total N 180 g•m−3 (105:75, NO3:NH4), P2O5 210 g•m−3, K2O 360 g•m−3, dry matter 37% (of which 82% organic matter), and water retention of 6 mL•g−1 dry matter.

The soil was contaminated by Cd(NO3)2‧4H2O (CAS N. 10022-68-1, Carlo Erba Reagents, Milan, Italy), K2Cr2O7 (CAS N. 7778-50-9, Carlo Erba Reagents, Milan, Italy), Cu(SO4)‧5H2O (CAS N. 7758-99-8, Prolabo, Paris, France), Pb(CH3COO)2 (CAS N. 301-04-2, Carlo Erba Reagents, Milan, Italy), Ni(NO3)2‧6H2O (CAS N. 13478-00-7, Carlo Erba Reagents, Milan, Italy), Zn(CH3COO)2‧2H2O (CAS N. 5970-45-6, Carlo Erba Reagents, Milan, Italy).

Hydrochloric acid (HCl 37%, Merck, Darmstadt, Germany), nitric acid (HNO3 65%, Merck, Darmstadt, Germany), double-distilled water (DDW ≥ 99.5%), standard multi-elemental reference solution (Ag, Al, Ba, Be, Bi, Ca, Cd, Co, Cr, Cu, Fe, K, Mg, Mn, Mo, Na, Ni, Pb, Sr, Tl, V, Zn [10 mg/l]) were used for heavy metals analysis by ICP-AES.

Algal strain and growth conditions

An Arthrospira platensis strain was cultivated by ApuliaKundi S.r.l. in open ponds (3,000 L per pond) under controlled greenhouse conditions with natural light at a temperature ranging from 22 to 28 °C.

The production cycle was monitored by turbidity measurements through a Secchi disk (200 mm diameter, Scubla, Remanzacco (UD), Italy). Once the disk was lowered into the algae suspension at 5÷6 cm and could not be seen anymore, the algae were collected. Overall, the production cycle lasted approximately three days. The collected algal biomass was filtered (40 μm filter), extruded, and cold dried at a temperature value lower than 38 °C. The study was conducted using the dried biomass in powder form, commonly named "spirulina."

Cultivation of Cannabis sativa L. (cv. Kompolti)

The cultivation of hemp by Enjoy Farm Soc. Coop. (Bitetto, Bari, Italy) took place indoors with controlled light and temperature conditions using a vegetative photoperiod of 18/6 h of light/dark for the first nine weeks, and a flowering photoperiod of 12/12 h of light/dark until harvest. (De Backer et al., 2012) Germination occurred 10 days after the seed sowing. After 50 days from germination, six plants were transplanted into 15 L pots filled with four kilograms of 75-days-old contaminated soil, in addition to two control plants transplanted into pots filled with four kilograms of uncontaminated soil. The pre-contaminated soil was prepared upon controlled addition into the soil of six different heavy metals, namely cadmium (Cd), chromium (Cr), copper (Cu), lead (Pb), nickel (Ni), and zinc (Zn). The choice of heavy metals and quantities used for contamination was based on the typical content of metals detected in different sites of the Apulia region of southern Italy, as provided by Agenzia Regionale per la Prevenzione e la Protezione dell'Ambiente (ARPA Puglia) and Acquedotto Pugliese S.p.A. Specifically, the pre-contaminated soil contained 16, 527, 200, 434, 357, and 448 mg•kg−1 of Cd(NO3)24H2O, K2Cr2O7, Cu(SO4)2‧5H2O, Pb(CH3COO)2, Ni(NO3)2‧6H2O, and Zn(CH3COO)2‧2H2O, respectively. Commercial uncontaminated soil (24 kilograms) was treated with 2.0 L of such salts aqueous solutions through four subsequent additions. The resulted mixture was stirred in a concrete mixer for three hours. Finally, the obtained contaminated soil was distributed into six pots (15 L each) and stored in the greenhouse under controlled conditions for 75 days.

Moreover, four out of the six Cannabis sativa L. plants grown in pre-contaminated soil received irrigation water (0.50 L of water × 56 times) containing spirulina (added in powder form). As a proof of concept, two different concentrations were tested: 1.0 and 0.50 g•L−1. Overall, each experimental condition was tested in duplicate, involving a total of eight hemp plants (Table 1).

Table 1. Description of the experimental conditions indicating the corresponding strategies of control, as well as pre-contamination soil in presence and absence of spirulina.
Experimental condition Description Plant ID
Control Two plants are grown on uncontaminated soil as a reference Ctrl
Pre-contaminated soil Two plants are grown on pre-contaminated soil C
Pre-contaminated soil + spirulina Two plants are grown on pre-contaminated soil with the addition of spirulina 1.0 g•L−1 CS1
Two plants are grown on pre-contaminated soil with the addition of spirulina 0.50 g•L−1 CS0.5

Plants were constantly monitored until the harvest (over two months after the sowing) by recording three main morphometric parameters: (i) plant height (cm), i.e., height of stem from ground to apex; (ii) the number of leaf stages; and (iii) drum diameter (mm) (see ESI for the full list of the morphometric parameters, Table S1). The biomass (g) of stem and leaf samples was determined upon lyophilization; the sum of the biomass of the stem and leaf samples collected from the same plant was computed too and is herein referred to as “plant biomass.” The residual amount of heavy metals (mg•kg−1) contained in leaves, stems, and soil after the harvest was measured; for each plant, also the sum of the heavy metal content of leaves plus stems was calculated (see ESI for further details, Table S2). The shoots of plants were collected and transferred in refrigerated packaging with dry ice from the greenhouse to laboratories for further analyses. Samples of soil from each pot were collected after harvesting and kept in plastic bags at room temperature until analysis. Leaves and flowers were separated from stems, and both portions of the plant were firstly freeze-dried at –50 °C and 0.180 mbar for 72 hours in a lyophilizer (Christ Alpha 1–4 LSC, Martin-Christ Gefriertrocknungsanlagen GmbH, Osterode am Harz, Germany). Then, the dried samples were mechanically ground in a blender, sieved through a mesh with a pore size of 0.5 mm, and stored at –20 °C until analysis.

Sample preparation and NMR analysis

An aliquot of 50 mg of each powdered sample was dissolved in 1.5 mL of buffer solution [Na2C2O4 (0.25 M), NaN3 (2.5 mm)] at pH 4.2 [pH adjusted by addition of HCl (37%)], sonicated for 10 minutes at 40 kHz, and vortexed (Advanced Vortex Mixer ZX3, VELP Scientifica Srl, Italy) for five minutes at 2,500 rpm. The mixture was transferred in a centrifuge tube (2 mL Nonsterile Centrifugal Filters, ThermoFisher Scientific) equipped with a polytetrafluoroethylene (PTFE) filter (0.2 µm) and centrifuged for 15 minute at 4,700 rpm (ROTOFIX 32 A, Hettich, Italy). The filtrate was used to fill the NMR tube.

NMR tubes were filled in by an automated system for liquid handling (SamplePro Tube, Bruker BioSpin) according to the following method: 630 µL of the obtained extract and 70 µL of TSP/D2O solution [3-(trimethylsilyl) propionic-2,2,3,3-d4 acid sodium salt in D2O (0.2% w/w)]. Three replicates were prepared for each sample.

One-dimensional 1H NOESY NMR spectra were recorded through a Bruker Avance 400 MHz spectrometer equipped with a 5 mm inverse probe and with an autosampler. The following acquisition parameters were used: pulse program = noesygppr1d; size of fid (TD) = 65 K; spectral width (SW) = 20 ppm; transmitter offset = 4.70 ppm; 90 ° hard pulse (p1) = 12.50 μs; power level for pre-saturation (pl9) = 59.10 dB; dummy scans (ds) = 4; number of scans (ns) = 128; acquisition time = 4.09 seconds; mixing time (d8) = 0.01 seconds; recycle delay (d1) = 5 seconds. Each spectrum was acquired using TOPSPIN 2.1 software (Bruker BioSpin GmbH, Rheinstetten, Germany) under an automatic process that lasted ca. 22 minutes and encompassed sample loading, temperature stabilization for five minutes, tuning, matching, shimming, and 90 ° pulse calibration.

NMR raw data (Free Induction Decays, FIDs) were processed using the software MestReNova 11.0 (Mestrelab Research SL, Santiago de Compostela, Spain). The FIDs relative to 48 1D 1H NOESY NMR experiments (48 experiments composed as follows: 8 leaf samples × 3 replicates, 8 stem samples × 3 replicates) were zero-filled with 128 K number of points and then underwent to Fourier transformation by applying an exponential multiplication function with a line broadening of 0.1 Hz. Phase and baseline were automatically corrected, and the TSP singlet signal set at δ = 0.00 ppm was used as a chemical shift reference.

ICP-AES analysis for quantification of heavy metals

Quantitative analysis of metals in samples of soil, stem powder, and leaf powder was carried out through ICP-AES, in three technical replicates. ICP-AES analyses were carried out at SAMER (Special Agency of the Chamber of Commerce of Bari). Samples of 200 mg were weighed into 100 mL digestion vessels using bi-distilled water, followed by acid digestion using 6 mL of 37% HCl (Merck, Darmstadt, Germany) and 2 mL of 65% HNO3 (Merck, Darmstadt, Germany). Digestion vessels were closed and placed into CEM Mars 1 microwave oven (CEM Corporation, Matthews, NC, USA). Power and time parameters used in the present work are part of an optimized library program for the analysis of trace metals in soil and plant material. The following digestion program has been, thus, selected for our purpose: 250 W (for two minutes), 0 W (for two minutes), 250 W (for five minutes), 400 W (for five minutes), and 500 W (for five minutes). After cooling to room temperature, the samples were filtered through 0.45 µm porosity filters and brought to 10 mL volume using bi-distilled water.

The quantitative measures of the metals were performed by ICP-AES (iCAP 6300 Duo, Thermo Fisher Scientific, Bremen, Germany). Quantities were reported in mg•kg−1. The amounts of Cd, Cr, Cu, Ni, Pb, and Zn were determined according to the UNI EN ISO 13657:2004 and UNI EN ISO 11885:2009 standards. The working wavelengths of the analyzed elements were 226.5 nm (Cd), 267.7 nm (Cr), 324.7 nm (Cu), 231.6 nm (Ni), 220.3 nm (Pb), and 213.8 nm (Zn). A concentrated standard reference solution was used at a concentration of 10 mg•kg−1 of Cd, Cr, Cu, Ni, Pb, and Zn. The six elements indicated, taken from 1000 mg•kg−1 solutions, were brought in a single solution from 100 mg•kg−1. Dilutions of 1, 0.5, 0.1, 0.01, 0.005, and 0 mg•kg−1 of the concentrated standard reference solutions were made in final volumes of 100 mL using HNO3 and HCl.

A calibration blank was used to obtain the analytical curve and was prepared by acidifying water with a mixture of 65% HNO3 and 37% HCl in a way that the standard and the sample reached the same value of acidity. Furthermore, the method blank contained all reagents in the same volumes used in the sample preparation and was used to identify possible contamination resulting from the acidic reagents used or equipment used during the sample preparation, including the filtration step. The calibration standard analytical curve was made of 1.0, 0.50, 0.10, 0.010, 0.0050, and 0 mg•kg−−1.

Statistical elaboration of NMR and ICP-AES data

The 48 processed 1D 1H NOESY spectra (8 leaf samples × 3 replicates, 8 stem samples × 3 replicates) were reduced to a numerical matrix (bucket-table) manageable for multivariate statistical analysis. The bucket-table was obtained by dividing the entire spectrum in the range of [-0.514, 10.486] ppm into 249 rectangular intervals (buckets) of 0.04 ppm in width, excluding the region [5.126, 4.566] ppm corresponding to the residual water signal. The underlying area of each bucket was normalized to the total intensity. The bucket table was imported into SIMCA 13.0.3 software (Umetrics, Umea, Sweden) to perform multivariate statistical analyses (MVA). The coordinates of the observations in the new space are called "scores," and the weight of the original variables on each PC are called "loadings." In the present study, the NMR spectra constituted the observations and the buckets constituted the x-variables. Buckets were centered and subjected to Pareto scaling (eachxj -variable was scaled to 1/sqrt(sdj), where sdj is the standard deviation ofxj -variable computed around the mean) to avoid noise inflation. The buckets in the range [-0.514, 0.486] ppm, including the TSP singlet, were excluded from statistical analysis.

Data from ICP-AES analysis and morphometric data were also included in statistical analysis as y-variables and were scaled to unit variance (UV), as recommended for variables expressed in different units (for eachyj -variable the base weight is computed as 1/sdj, where sdj is the standard deviation of yj-variable computed around the mean). Principal component analysis (PCA) and orthogonal partial least squares analysis (OPLS) were performed. The quality of statistical models was assessed based on the parameters R2 and Q2, which represent the descriptiveness (goodness-of-fit) and the predictivity in cross-validation (goodness-of-prediction in cross-validation), respectively. For OPLS models, permutation tests were carried out to verify the absence of model overfitting.

Results and discussion

NMR analysis (1D 1H NOESY experiments) allowed for the identification of a pool of metabolites contained in typical aqueous extracts (pH = 4.2) of leaves (Fig. 1a) and stems (Fig. 1b) collected from a plant of hemp cultivated under controlled conditions in uncontaminated soil (see ESI for the full list of metabolites, Table S3). The main representative classes of metabolites included organic acids (i.e., lactic, citric, tartaric, fumaric, formic, and acetic acids), amino acids (i.e., alanine, asparagine, γ-aminobutyric acid, arginine, glutamine, isoleucine, proline, tyrosine, leucine), and carbohydrates (i.e., β-glucose, α-glucose, fructose, α-galactose, β-galactose). Moreover, 1D 1H NOESY spectra also contained signals related to trigonelline, choline, and ethanol.


Fig1 Musio EnviroAdv2022 7.jpg

Fig. 1 Typical 1D 1H NOESY spectra (Bruker Avance 400 MHz, D2O) of aqueous extracts of leaves (a.) and stems (b.) of Cannabis sativa L. The spectral region containing the residual water signal (4.78 ppm) was hidden. The spectral regions included in rectangles were expanded for clarity.

A holistic approach was applied during the present study by combining the information provided by the 1D 1H NOESY experiments, the ICP-AES quantitative analysis of residual metal, and the data derived from the morphometric inspection.

As a first task, the relation between the variations in the metabolic composition and the calculated amounts of residual heavy metals in the plant was considered. Thus, the OPLS analysis was performed considering the spectral buckets as x-variables, and the metal content and plant biomass as y-variables (see ESI for OPLS model parameters, Table S4). Such a model was characterized by five predictive components P1-P5 that explained 59.5% of x-variance (R2X[cum] = 0.595), 92.2% of y-variance (R2Y[cum] = 0.922) and 87.6% of y-variance modelled by x-variables (R2 = 0.876), and eight orthogonal components O1-O8 that explained about 38% of the x-variance unrelated to y-variance (R2X[cum] = 0.383). Two main separate clusters of observations were noticeable along P1. They corresponded to the two parts of the plant, namely the leaves (LE) and the stems (ST) (Fig. 2a, LE vs ST). Inside each group, two evident subgroups were distinguishable along P2, which were identified as the vigorous plants (V) and the weak ones (W) (Fig. 2b, V vs W). The components P1 and P2 explained 41 and 9.8% of the information in x-space (R2X[P1] = 0.41 and R2X[P2] = 0.098), suggesting an important variability of the metabolic composition along with the different tissues of the plant and according to the wellness of the plant. Indeed, the same components P1 and P2 explained 48.5 and 15.2% of the information in y-space (R2Y[P1] = 0.485 and R2Y[P2] = 0.152), suggesting a differential distribution of the y-variables, namely the metals concentration and the plant biomass, according to the different tissue of the plant and the health status of the plant (Fig. 2c). The visual inspection of the original harvested plants (see ESI for further details, Fig. S1), combined with the plant biomass values, suggested that the samples included in the subgroup with t[2]<0 (Fig. 2b, red scores) corresponded to the less healthy plants (weak, W), whereas the observations included in the subgroup with t[2]>0 corresponded to the more vigorous plants (V, Fig. 2b, yellow scores). The analysis of the loading plot (pq[1] vs pq[2]) allowed the identification of the most significant variables towards the distribution of the scores, including the metal concentration, biomass amount, and spectral regions (Fig. 2c).


Fig2 Musio EnviroAdv2022 7.jpg

Fig. 2 OPLS-DA applied to spectral data (x-variables) and metal content and plant biomass data (y-variables). (a.) Scores plot displaying t[1] vs t[2], the observations are colored according to the tissues of the plants: leaves (green circle, LE), stems (blue square, ST). (b.) Scores plot t[1] vs t[2], the observations are colored according to the health status of the plants: vigorous (yellow circle, V), weak (red square, W). (c.) Loading plot displaying the relationship between the x-variables (spectral regions) and the y-variables (metal content and plant biomass) for the first and second predictive components pq[1] vs pq[2] (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.).


References

Notes

This presentation is faithful to the original, with only a few minor changes to presentation. Some grammar and punctuation was cleaned up to improve readability. In some cases important information was missing from the references, and that information was added.