Jing
Li
ab,
Hae Yoon
Cho
b,
Sung Won
Kwon
b and
Seul Ji
Lee
*b
aCollege of Life Science, Shanghai Normal University, Shanghai 220234, China
bCollege of Pharmacy, Seoul National University, Seoul 08826, Republic of Korea. E-mail: dltmfwl2@snu.ac.kr
First published on 26th February 2021
Many natural products have demonstrated functionality as novel, green sorbents for organic compounds. However, only limited reports exist on the use of such green materials as solid-phase extraction (SPE) sorbents for select organic acids. In this study, we employed pollen grains as a hydrophilic sorbent and investigated the influence of various extraction parameters using a series of experimental designs. The chemical structure and surface properties of the prepared sorbent were investigated by Fourier-transform infrared spectroscopy and scanning electron microscopy. The Plackett–Burman design was used to experimentally screen for parameters that significantly influenced the extraction performance. Three selected parameters were then statistically optimized by applying a central composite design combined with a response surface methodology. Phenolic acid residues were determined and quantified using high performance liquid chromatography with ultraviolet detection; a mass spectrometric detector in the selected ion monitoring mode was also used for identification. As a practical example, phenolic acids in the soil were successfully separated by the developed pollen sorbent. These results therefore indicate that pollen grains can be considered as a sustainable, green, and safe alternative to bare silica for extraction and separation applications.
In the majority of studies carried out thus far, sample preparation plays an important role in obtaining reliable high-quality data by enriching the analytes and reducing matrix effects in the subsequent steps. Although liquid–liquid extraction is a core technology in this regard, conventional liquid–liquid extraction processes usually utilize solvents that are flammable, volatile, or toxic.7 Thus, since their introduction in 1978, disposable cartridges for solid-phase extraction (SPE) have become popular for a wide variety of matrices, including urine, blood, water, beverages, soil, and animal tissue.8,9 In recent years, significant efforts have been directed toward developing new, advanced sorbents with improved selectivities (i.e., improved clean-up effectiveness) and sorptive capacities (i.e., enhanced pre-concentration factors).10 The most common sorbents include surfactant-modified sorbents,11 mixed-mode polymeric sorbents,12 and molecularly imprinted sorbents,13 all of which have been synthesized using conventional chemicals.
For example, silica gel chemically bonded with various organic compounds has received considerable attention as a sorbent due to its good mechanical and thermal stabilities.14,15 However, silica is mainly obtained by mining and purifying quartz, and the chemicals used in these processes, in addition to the produced silica dust itself, can directly harm humans by causing silicosis, cancer, and an increased risk of tuberculosis.16,17 Another issue is the related, negative environmental impact from the generation and disposal of industrial wastes and emissions,18 which remains challenging to reduce despite stringent government regulations.
As a natural green material, pollen grain has been applied in drug delivery,19 microsphere technologies,20,21 and template materials.22 Its most striking structural feature is its tough outer coating (known as the exine), which is formed by an extremely stable and complex biopolymer called sporopollenin and is highly resistant to chemical attack and high temperatures.23–25 In addition, the apertural furrows and hydroxyl groups present on the pollen surface provide it with a strong adsorption ability, thereby rendering the pollen grain a promising hydrophilic SPE sorbent for miniaturized methodologies in green analytical chemistry. Moreover, the many reactive sites on the pollen surface, including the hydroxyl groups, offer opportunities for surface functionalization to enhance the compatibility between the pollen and the polymer matrix, although at present, no studies have explored this idea.26 Furthermore, pollen grains from different plants exhibit wide variations in size (5–200 μm diameters) and shapes (round, elliptical, and multifaceted), which are advantageous in the context of absorptive materials. The performances of pollen-based sorbents in extraction and separation processes can also be tailored by changing the proportions of water and acids in the activation solvent or eluting solvent. Nevertheless, few studies have reported the use of pollen grains for extraction purposes.
For the purpose of this study, we employed phenolic acids as model analytes because they are one of the most common types of organic acids. Moreover, phenolic acids bear varying numbers of hydroxyl groups on their aromatic ring, making them very polar compounds. Therefore, if the developed, hydrophilic SPE cartridges can successfully extract phenolic acids, this method will likely be effective for most polar molecules. Furthermore, phenolic acids have been found in root exudates, which have been reported to aid in amending micro biomass and cycling organic matter.27–31 In addition, it has been found that varying quantities of the different phenolic acids in the soil evokes specific microbial responses in terms of the biomass, activity, and community composition. This in turn can inhibit seed germination and plant growth.32 These chemicals are also found in significantly higher levels in soils that have been subjected to extensive cropping compared to those that have undergone normal cropping. Importantly, few studies have reported the accurate quantification of organic acids in soil through the use of sensitive, eco-friendly, and low-cost methods.
In this study, the pollen grains predominantly interact with the phenolic acid analytes via hydrogen bonding through their surface hydroxyl groups. Although a few studies have reported the application of pollen grains in SPE,33–36 the optimization of the extraction conditions was incomplete; thus, sorbent parameters that have the strongest influence on the extraction efficiency were not identified. To address these issues regarding hydrophilic interaction liquid chromatography-SPE (HILIC-SPE), we employed the Plackett–Burman factorial design to identify the significant experimental variables affecting the extraction efficiency. Thereafter, the extraction process was optimized using a central composite design (CCD) approach. Finally, the applicability of the proposed method toward phenolic acids in soil was assessed. Compared with conventional sorbents for HILIC, pollen grains are expected to be eco-friendly and inexpensive. Moreover, they offer a large selection of particle sizes (5–200 μm) and shapes for diverse applications, thereby rendering them a promising class of green, absorptive materials for use in analytical chemistry.
For all experiments, analytical, reagent-grade chemicals and solvents were used. Seven of the phenolic acids (gallic acid, 4-hydroxybenzoic acid, vanillic acid, caffeic acid, p-coumaric acid, ferulic acid, and sinapic acid) were purchased from Sigma-Aldrich (St. Louis, MO, USA). Protocatechuic acid, trans-m-coumaric acid, 1-undecanol, and 2-dodecanol were obtained from TCI (Tokyo, Japan). Acetone, water, methanol, and ACN of LC grade were obtained from Duksan Chemical Co. Ltd. (Ansan, Korea). All other reagents were purchased from Sigma-Aldrich unless otherwise noted.
The following commercial cartridges were used for SPE: Oasis HLB (3 mL) from Waters; discovery DSC-18 (3 mL) from Supelco; Strata-X 33u polymeric reversed-phase (3 mL) from Phenomenex; and silica (3 mL), C18 (3 mL), and SAX (3 mL) from Alltech.
High-performance liquid chromatography (HPLC) analysis was conducted on a Flexar FX-10 system (PerkinElmer, Shelton, CT, USA) with a Flexar FX photodiode array (PDA) (PerkinElmer, USA). The detection wavelengths were 254, 280, and 310 nm. An Agilent poroshell EC-C18 column (2.1 × 150 mm) packed with 2.7 μm core–shell particles was used for chromatographic separation of the nine phenolic acids to improve the resolution with shorter retention times, higher sensitivities, and superior performances.
Design-Expert software (Version 9.0, Stat-Ease Inc., Silicon Valley, CA, USA) was used for experiment design and data analysis.
A thin polyethylene frit was placed at the bottom of a 3 mL hollow SPE cartridge. The prepared pollen grain sorbent (650 mg) was accurately weighed in the laboratory and packed into the cartridge. The sorbent was then covered by another polyethylene frit and compacted at this thickness to ensure a smooth and flat surface.
The same SPE procedure was followed for all sorbents. Initially, the cartridges were preconditioned with water/ACN (20/80, v/v, 2 mL) and activated by ACN (2 mL). Thereafter, an aliquot (800 μL) of the prepared sample solution was loaded onto the pollen grain SPE cartridge. After washing the cartridge with ACN (1 mL), a mixture of water/ACN/FA (30/65.5/4.5, v/v/v, 3 mL) was used for elution. The eluate was collected in an Eppendorf tube and evaporated to dryness. The residue was dissolved in water (200 μL) and filtered through a 0.2 μm membrane filter (Whatman, Piscataway, NJ, USA) prior to injection.
Code | Factor (unit) | (−1) value | (+1) value |
---|---|---|---|
A | Sorbent amount (mg) | 200 | 600 |
B | Activation solvent volume (mL) | 2 | 3 |
C | Sample volume (mL) | 1 | 2 |
D | Clean-up solvent type | Acetonitrile | Acetone |
E | Clean-up solvent volume (mL) | 1 | 2 |
F | Proportion of formic acid in eluting solvent (%) | 1 | 3 |
G | Proportion of water in eluting solvent (%) | 30 | 40 |
H | Eluting solvent volume (mL) | 2 | 3 |
To determine the optimal extraction conditions, three factors were performed with six star points each placed at a distance of α from the central point. Twenty experiments were required, including six central points, and they were conducted randomly. The conditions set for each experiment are listed in Table S1, ESI.† Moreover, the profile for the predicted values and the desirability option was used to optimize the extraction process.
The surface structures of the pine pollen grains were visualized using SEM. Prior to surface treatment (Fig. 1a), the apertural furrows and longitudinal folds in the exine were visible. After refluxing in methanol and drying in an oven, the exine was all that remained, and its apertural furrows became more prominent (Fig. 1b). From the SEM images, the pollen grains were found to possess an average diameter of 30–50 μm.
![]() | ||
Fig. 1 Scanning electron microscopy (SEM) images of the pollen samples (a) before treatment and (b) after refluxing in methanol. |
The diffused reflectance FT-IR spectrum of the pollen grains is shown in Fig. S2.† As indicated, a strong and broad band was visible at 3300 cm−1, which is characteristic of –OH stretching vibrations. Relatively weaker bands were also observed at 2920 and 2851 cm−1, corresponding to the –CH stretches of –CH2. Furthermore, the absorption band at 1028 cm−1 is characteristic of C–O stretching vibrations, revealing the presence of the C–OH functionality (Fig. 2).
![]() | ||
Fig. 2 Characterization of the pollen grain surface by diffused reflectance Fourier-transform infrared (FT-IR) spectroscopy. |
Analysis of variance was performed to examine whether the studied experimental factors (Table 1) were significant for the performance of the proposed method. These eight factors were assessed at two levels coded as + (high) or − (low). Their effects were calculated based on the peak areas averaged over all the analytes in the chromatograms, and the significance was evaluated using Pareto charts (Fig. 3). The results were found to agree with those from the regression line plotted in the Design Expert program. For an analyte, a “probability > F′′ value < 0.05 indicates that the model and the term are significant. In the studied experimental domain, a positive sign indicates that increasing the variable will enhance the signal (i.e., the average peak areas of all analytes). Based on the obtained results, we further optimized three of the parameters (i.e., the sorbent amount, the sample volume, and the proportion of FA in the eluting solvent) using CCD, while the other five parameters (i.e., the activation solvent volume, the clean-up solvent type, the clean-up solvent volume, the proportion of H2O in the eluting solvent, and the eluting solvent volume) were kept constant.
The five fixed parameters were as follows: activation solvent volume = 2 mL, clean-up solvent type = ACN, clean-up solvent volume = 1 mL, proportion of H2O in the eluting solvent = 30%, and eluting solvent volume = 3 mL. Meanwhile, the other three parameters were varied in the following ranges during optimization using response surface methodology (RSM): sorbent amount = 200–800 mg, sample volume = 0.5–1.5 mL, and proportion of FA in the eluting solvent = 0–6%. The experimental order and the levels of the coded factors and variables are summarized in Table S1 (ESI†).
Three response surfaces were obtained from the CCD results for each phenolic acid. Using gallic acid as a typical example, the response surfaces and the corresponding contours are illustrated in Fig. 4. For each phenolic acid, a different quadratic multiple regression model was obtained for the variables and the response. The main effects, the interaction effects, and the quadratic effects were evaluated through analysis of variance at the 95% confidence level (p < 0.05). Moreover, a lack of fit, which measures the failure of the model to represent data in the experimental domain at points not included in the regression, was also checked. It was shown to be not significant relative to the pure error, indicating a good response to the model.
The models were expressed as second-order polynomial equations for the extraction yield (Y) and the coded factors as follows:
Ygallic acid = 380![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Yprotocatechuic acid = 606![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Yp-hydroxybenzoic acid = 1![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Yvanillic acid = 977![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Ycaffeic acid = 1![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Yp-coumaric acid = 1![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Yferulic acid = 631![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Ysinapic acid = 409![]() ![]() ![]() ![]() ![]() ![]() |
Ym-coumaric acid = 2![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
The regression coefficients (R2) for gallic acid, protocatechuic acid, 4-hydroxybenzoic acid, vanillic acid, caffeic acid, p-coumaric acid, ferulic acid, sinapic acid, and m-coumaric acid were determined to be 0.7376, 0.7535, 0.9201, 0.9507, 0.9313, 0.9174, 0.9327, 0.9240, and 0.9081, respectively, which are in close agreement with the experimental results. Each graph shown in Fig. 4 plots the effect of two variables, while the third variable is maintained constant at the center point. From Fig. 4a, the extraction yield (peak area) is large in the case of abundant pollen grains or when a large amount of pollen grain was used for adsorption. These observations can be explained by the fact that in the absence of pollen grains, the ratio of the sorptive area to the available analytes is low; thus, there is a lower chance for analyte removal. Fig. 4b shows the effect of the pollen grain amount and the proportion of FA in the eluting solvent. As indicated, the response increased upon increasing the FA proportion when the sample volume was fixed, and the amount of pollen grain was ≤650 mg. In HILIC-SPE elution, both the sorbent and the analytes possess positive charges under acidic conditions.38 Therefore, a more acidic solvent facilitates analyte elution. As shown in Fig. 4c, when a smaller amount of sample was introduced, the extraction efficiency increased irrespective of the FA proportion.
Design Expert software was then used to maximize the average response of the target analytes, and the optimized model variables were determined to be as follows: sorbent amount = 650 mg, sample volume = 0.795 mL, and proportion of FA in the eluting solvent = 4.5%. All subsequent experiments were performed using these conditions.
The linear ranges were 10–40 μg mL−1 for gallic acid, 0.4–40 μg mL−1 for protocatechuic acid, 10–40 μg mL−1 for 4-hydroxybenzoic acid, 10–40 μg mL−1 for vanillic acid, 10–40 μg mL−1 for caffeic acid, 0.2–20 μg mL−1 for p-coumaric acid, 0.2–20 μg mL−1 for ferulic acid, 0.4–40 μg mL−1 for sinapic acid, and 0.2–20 μg mL−1 for m-coumaric acid. The method precision was below 11.57% RSD, and the relative recovery was between 77.58% and 118.98% in all concentration ranges (Tables S2 and S3, ESI†).
To isolate the phenolic acids, the C18 column relies on van der Waals forces, hydrogen bonds, or dipole–dipole interactions. Polymeric sorbents are stable over a broader pH range and possess a larger surface area, and are advantageous because of the selective π–π interactions with aromatic-containing analytes. The recoveries of the various phenolic acids achieved using the pollen grain SPE and the commercial SPE sorbents are shown in Fig. 3. As indicated, the pollen grain sorbent was found to be appropriate for the isolation and preconcentration of polar compounds. As an example, gallic acid is more polar than the other studied compounds, and so its recovery was ∼75% for pollen grain but <40% for the C18 sorbent. Although some of the polymeric sorbents displayed higher recoveries than the pollen grain sorbent, the eco-friendliness and low cost of the latter still render pollen grains a promising sorbent (Fig. 5).
To measure the bed capacity, aliquots (800 μL) of the nine standard solutions (10–10000 μg mL−1) were loaded onto the SPE sorbent (650 mg) conditioned with water/ACN (20/80, v/v, 2 mL) and ACN (2 mL). The cartridge was then washed with ACN (1 mL), and the analytes were eluted using water/ACN/FA (30/65.5/4.5, v/v/v, 3 mL). The eluate was collected and evaporated to dryness under a stream of nitrogen. Finally, the residue was dissolved in methanol (200 μL) and analyzed by HPLC-UV.
Recovery is commonly determined as the mass ratio between the analyte in the eluent and that introduced to the sorbent bed. Here, we represent the bed capacity by the absolute recovery at different analyte concentrations (Fig. 6). We observed that the bed capability is related to the molecular polarity of the sorbent, and the recovery is enhanced when the polarity is increased. In addition, we found that the bed capability decreased considerably when the analyte concentration was in the range of 0–1000 μg mL−1, and then remained stable up to 5000 μg mL−1.
![]() | ||
Fig. 6 Bed capacity curves for pollen grain sorbents, determined by the absolute recovery and analyte concentration. |
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/d1ay00044f |
This journal is © The Royal Society of Chemistry 2021 |