Bimetallic nanocolloidal plasmonic array for polyphenols characterization and calibration-free antioxidant capacity evaluation

doi:10.21203/rs.3.rs-4842921/v1

Download PDF

Research Article

Bimetallic nanocolloidal plasmonic array for polyphenols characterization and calibration-free antioxidant capacity evaluation

https://doi.org/10.21203/rs.3.rs-4842921/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Phenolic compounds (PC) are natural molecules with antioxidant capacity (AoC) with recognized technological and health properties; rapid and easy-to-use approaches for PC class and AoC evaluation are highly required. Herein, a bimetallic plasmonic nanoparticles-based approach for the untargeted evaluation of PC-pattern and AoC is proposed. The rationale relies on the PC's ability to drive the formation of bimetallic silver/gold nanocolloidal ‘probes’ with different conformations. Ag/Au bimetallic nanostructures, according to the PCs’ amount and class, return characteristic plasmonic and colorimetric tags. Plasmonic indexes are proposed to assess the dominant PC classes, while the colorimetric response, analyzed simply by a smartphone, is employed to obtain an AoC score, with no calibration. The methods were tested with PCs belonging to different chemical classes, and challenged to classify different food samples; the proposed approach allows PC-dominant class identification and AoC-evaluation consistent with HPLC-MS/MS and conventional photometric assays.

colorimetric assay

plasmonic

antioxidant capacity

polyphenols clustering

nanocolloids

food samples

Metal nanoparticles (MNPs) optical features, related to their ability to interact with electromagnetic radiation in the Uv-Visible region, have been widely investigated by the scientific community and exploited in different fields [1]. These features are mainly attributed to the Localized Surface Plasmon Resonance phenomenon (LSPR), which consists in the collective oscillation of conductive electrons localized onto the MNP surface triggered by incident electromagnetic radiation [1]. LSPR is structure-dependent and varies according to MNP size, morphology, and surface chemistry (i.e., metal nature, surface functionalization, etc.); it is also influenced by the chemistry of the dispersing medium [1, 2]. This LSPR-tunability allows customization of the MNP properties (optical band, absorbance peak profile/shape, signal intensities, etc.) to fulfill the required need for the final application [3].

Among MNPs, gold and silver NPs (AuNPs and AgNPs, respectively), thanks to their high molar extinction coefficient, resulting in prominent plasmonic signals, and their ease of synthetisis route and colloidal stability, have received enormous scientific attention [1]; they have been largely used as plasmonic sensing elements, colorimetric-tags, and signal enhancers in several analytical applications, including SERS, Raman, and colloidal spectrophotometric-assays [1]. MNPs-based optical sensing strategies are mainly devoted to quantitative analysis, where the analytical signal is based on the absorbance intensity variation (increase or decrease) and/or LSPR band-shifting (blue/red shift) induced by the target analyte [4, 5]. In this framework, there are no methods devoted to relating the MNPs' plasmonic profile with quali-quantitative analytical information.

In addition to methods based on plasmonic output, the opto-colorimetric properties of MNP pushed to colorimetric (bio)sensing strategies development, exploiting them as colorimetric reporters/tags; this allows to simplify the analytical detection approach allowing simple naked eye or smartphone-based readout [1, 6]. In these methods, there is always an LSPR change, where LSPR-active MNPs, depending on their excitation frequency, return visible colors [1, 7].

Taking a look at the literature, several colorimetric assays based on single metal nanoparticles are reported; however, the use of MNP in sensor array formats is not largely explored [1, 6, 8, 9]. In these studies, the MNPs-based arrays are composed mainly of previously synthesized MNPs, which are used to obtain discriminatory color maps [6, 8]. Regarding the analysis of antioxidant compounds, an interesting approach was proposed by Bordbar et al. [6], which proposed a color fingerprinting of 20 food antioxidants using a set of Ag- and AuNPs characterized by different surface functionalization. Li et al. [8] were able to discriminate the antioxidant capacity (AoC) of biomolecules evaluating their ability to prevent the etching of triangular-AgNPs in the presence of oxidizing agents. Despite the wide interest in antioxidant compounds, and the opto-colorimetric MNP-based arrays' potentiality to return a pool of information beyond the mere quantification of an analyte, their use for antioxidants' quali-quantitative assessment of chemical classes and reactivity is still unexplored.

In this work, a plasmonic array based on the antioxidant compounds driven formation of gold/silver bimetallic nanoparticles (BiMNPs) is proposed; the array is based on different colloidal syntheses occurring simultaneously in ELISA plates, conceived to give rise to different BiMNPs formation routes. The analytical strategy relies on the ability of PCs to reduce the Au and Ag precursors (i.e. metal cations) in the form of BiMNPs with different final colloidal conformation (i.e., nano-sponge, core-shell structure, Ag/Au NPs coexisting population, etc.); the bimetallic arrangements returns different plasmons and colors that can be attributed to the different PCs structure and AoC. To understand the rationale behind this strategy, the BiMNPs were characterized microscopically (TEM) and spectroscopically (XPS and UV-Vis).

To improve the ability of the array to differentiate, for each bimetallic LSPR profile (corresponding to one reaction mix), four key absorbance values were selected and normalized to obtain plasmonic indexes. The final array consists of 16 outputs (4 reaction-mix x 4 plasmonic-indexes), that, via multivariate analysis allow the recognition of PC dominant structural class. In addition, a fast smartphone-based colorimetric strategy for AoC evaluation is also proposed. The obtained colorimetric patterns, elaborated in the RGB color space, return a quantitative score concerning the PC's AoC with no need for calibration.

The BiMNPs array has been tested with 9 selected PC standards and challenged for the classification of 13 food samples. Eventually, the ability to return qualitative information on the PC prevalent chemical class for food samples was confirmed via HPLC–MS/MS, while the reliability of the AoC evaluation was proved by comparing the data obtained with conventional photometric assays (i.e., ABTS, Folin-Ciocalteu, and Aluminum Complex Formation).

2.1. Reagents and stock solutions

A detailed description is reported in section SI.2.1.

2.2. Phenolic compounds and samples

P-hydroxybenzoic acid (p-HB), p-coumaric acid (p-CM), naringenin (NR), caffeic acid (CF), rosmarinic acid (RM), catechin (CT), gallic acid (GL), ellagic acid (EL) and tannic acid (TN) were used as model phenolic compounds (PC). The complete list of the employed PC, the respective acronyms, and their chemical and sub-chemical classes are listed in Table S1.

Food samples were purchased in local markets; their names and acronyms are reported as follows: ground and whole cloves (Cl1 and Cl2, respectively), coffee powders of different brands (Cf1, Cf2, and Cf3), roasted coffee beans’ peel (Cf4), green teas of different brands (Gt1, Gt2, and Gt3), dried mint leaves and mint infuse (Mn1 and Mn2, respectively), dried sage (Sg) and dried thyme (Ty). Samples were extracted according to Della Pelle et al. [5]; in brief, 0.5 g of sample were extracted with 10 mL of MeOH:H₂O (80:20 v/v) via orbital stirring (200 rcf; 1h) in the dark. Sample extracts were centrifuged at 4000 g and the supernatants were collected and stored at − 20°C.

2.3. Apparatus

A detailed description is reported in section SI.2.3.

2.4 Bimetallic nanocolloids synthesis and plasmonic indexes

BiMNPs were produced using four reaction mixtures named mix 1 (M1), mix 2 (M2), mix 3 (M3), and mix 4 (M4) based on different Ag⁺ and Au³⁺ metal precursors ratios; BiMNPs synthesis was simultaneously performed in 96-well ELISA plates in a final reaction volume of 200 µL.

For each mix, 178 µL of 0.1 M phosphate buffer pH 9 (PB) was placed in the microwell, followed by 2 µL of cetyltrimethylammonium chloride solution (CTAC); then, 10 µL of Au³⁺ and Ag⁺ at different Au³⁺/Ag⁺ concentration ratios were added. Ratios were as follows: M1 = 500 µM / 250 µM, M2 = 500 µM / 500 µM, M3 = 500 µM / 750 µM, M4 = 250 µM / 500 µM. Eventually, 10 µL of properly diluted PC-standard/sample was added to each mix to trig the bimetallic colloid formation; the ELISA plate was orbitally shaken in the dark for 10 min at 300 rcf, then the reaction was stopped incubating at + 4°C for 10 min.

Four plasmonic indexes (PI) named λ₄₃₀, λ₅₁₀, λ₅₅₀, and λ₆₅₀ were proposed to interpret the bimetallic colloids' visible spectra. Each PI is a ratio of the absorbance (Abs) recorded at two wavelengths, aimed to maximize the spectral information; the indexes were as follows: λ₄₃₀ = Abs_λ430/ Abs_λ330, λ₅₁₀ = Abs_λ510/ Abs_λ550, λ₅₅₀ = Abs_λ550/ Abs_λ600, and λ₆₅₀ = Abs_λ650/ Abs_λ550.

2.5. Bimetallic nanocolloids physico-chemical characterization

A detailed description of the physico-chemical characterization performed on the bimetallic colloids is reported in section SI.2.5.

2.6. Phenolic compound analysis via conventional methods

Samples were analyzed using these analytical methods: (SI.2.6.1) phenolic pattern evaluation through HPLC–MS/MS targeted analysis; (SI.2.6.2) total phenols content evaluation through Folin-Ciocâlteu-based assay; (SI.2.6.3) antioxidant capacity evaluation through ABTS•⁺ radical-based assay; (SI.2.6.4) flavonoid evaluation through aluminum complexation-based assay.

A detailed description of the analysis performed is reported in section SI.2.6.

2.7. Phenolic compounds clustering and plasmonic indexes multivariate analysis

R software [10] was used for statistical analysis and data management. Three datasets were processed: (i) mass spectrometry dataset made by 13 observations (samples) and 31 variables (PCs identified by mass spectrometry); (ii) spectrophotometric dataset made by 9 observations (PC standards) and 16 variables (PI collected from the four reaction mixtures); (iii) spectrophotometric dataset made by 117 observations (samples extracts performed and analyzed in triplicate) and 16 variables (PI) collected from the four reaction mixtures). Before any statistical analysis, datasets were standardized using the Z-score transformation (n-1).

Heatmap was built using dataset-i with the standard parameters of the ‘heatmap’ command in R software computing the distance (dissimilarity) between both rows and columns with the function ‘dest’ and the agglomerative hierarchical clustering (AHC) with the function ‘hclust’. The cluster distance was computed using the Dunn index provided by clValid Package (default parameters) [11]. Statistical significance was assessed using analysis of variance (ANOVA) with the Tukey honestly significant difference (HSD) multiple comparison analysis using ‘stats’ R package. The criterion for statistical significance of differences was p-value < 0.05 for all the comparisons.

Multivariate statistical analysis was performed using principal component analysis (PCA) and partial least square discriminant analysis (PLS-DA) using the mdatools Package [12]. PCA was performed on dataset-ii and -iii.

PLS-DA was run on dataset-iii; in this case, the Y block was a categorical matrix of 117 rows and 4 columns related to the 4 different clusters found in mass spectrometry hierarchical analysis (cluster-a, -b, -c, and -d). The ‘Leave one out’ procedure was used for the cross-validation (CV) technique by dividing the dataset into CV groups equal to 1, removing them one at a time from the training set used to calibrate the PLS-DA classification model. To avoid overfitting, the PLS-DA was carried out using a k-fold cross-validation approach (k = 10) using the R package KODAMA with the function ‘pls.double.cv’ [13].

2.8. Smartphone-based antioxidant capacity score via colorimetric bimetallic array

Pictures of the BiMNPs array were collected in a black reading chamber (15×15×15 cm) with a smartphone camera (XIAOMI MiA3) at constant focus (15 cm of focus distance) in lighting conditions, using a white background.

BiMNPs color analysis was performed with ImageJ free software using the RGB color space. The RGB hue intensity was collected and converted into RGB% according to the following equation: $\:RGB\%=\:\raisebox{1ex}{$\left({RGB}_{sample}\times\:100\right)$}\!\left/\:\!\raisebox{-1ex}{$255$}\right.$.

Colorimetric data obtained for the four BiMNP synthesis mixtures (M1, M2, M3, M4; section 2.4) were interpreted using a score grid to define an antioxidant capacity total score (AoC T.S.). The bimetallic nanocolloidal plasmonic nanoparticles array (BiMNPs array) colorimetric response, employed to fill the score grid, is obtained directly in ELISA plates; PC standards and samples were analyzed according to section 2.4 at different fixed concentrations/dilutions.

Colorimetric quartiles definition. RGB% values (0-100%) obtained for each BiMNP were divided into four quartiles (Q), at each quartile was attributed a score and a colorimetric tag: Q1 RGB ≤ 25%, score = 1 (gray tag); Q2 25% ≤ RGB ≤ 50%, score = 2.5, (yellow tag); Q3 50% ≤ RGB ≤ 75%, score = 5 (red tag); Q4 RGB ≥ 100%, score = 10 (black tag).

AoC T.S. extrapolation. The BiMNPs array score grid was constructed by placing the scores, obtained from the different BiMNP synthesis mixtures (M1-M4) as rows; the columns report the scores obtained at different PC/sample concentrations/dilutions. For simplicity, in the score grid, the scores are placed as colorimetric tags. The sum of the scores, obtained from the single columns (named partial scores; P.S.), represents the AoC T.S.

Scheme 1. AoC T.S. scores and extrapolation. (A) Colorimetric quartiles: RGB% ranges, scores values and tags. (B) Example of the AoC T.S. extrapolation via BiMNPs array score grid for a generic sample; P.S. = partial score value, T.S. = total score value.

3.1. Bimetallic nanocolloids optical and morpho-chemical features

Bimetallic nanocolloid (BiMNPs) synthesis mediated by phenolic compounds (PCs) was systematically investigated to obtain the maximum information varying the reaction medium. Optimal pH and time were studied using catechin (CT) as model-PC using the reaction mix M2, following the procedure reported in section 2.4; CT was used since it is widely found in food and has a well-known ability to form MNPs [14]. The plasmonic signal was monitored at 430 and 550 nm, typical of the maximum LSPR of AgNPs and AuNPs, respectively [5].

The pH was studied between 6–12 (Fig. S1A). pH 9.0 was selected since represents the best compromise between LSPR peak intensity and reproducibility for both wavelengths; at this pH, PC’s hydroxylic groups are ionized and prone to easily donate electrons. pH values lower than 9.0 result in scarce BiMNPs production, whereas higher values induce casual and unreproducible synthesis [5, 15]. The BiMNPs reaction time has been also studied (Fig. S1B); 10 min resulted enough to achieve reproducible synthesis and stable colloids.

Figure 1A-C reports the colorimetric and LSPR profile of BiMNPs obtained, in the optimized conditions, for increasing amounts of p-coumaric acid (p-CM), rosmarinic acid (RM), and CT; these PC were selected as model molecules since belong to different chemical classes and possess different chemical structures, stereochemistry, and reducing phenolic moieties [16].

The reported spectra show the three different types of BiMNP nanostructures that can be obtained with different PCs (see below); they are achieved employing the reaction mix M1. In Fig. 1D-F, to better appreciate the differences between LSPR signatures and the LSPR maximum (λ_max) shifts, the first derivative was used to process the spectra [3, 17]; the ‘zero-crossing’ points coincide with the λ_max.

The PC chemistry drastically influences the BiMNPs synthesis; indeed, in the same conditions, different LSPR 'signatures' and colloid colors are obtained. In all the cases, the plasmonic signal intensity varies according to the PC concentration, while the color changes not only in intensity but even in hue depending on the nanostructure formed.

In detail, p-CM, a mono-phenol, induces the synthesis of BiMNPs characterized by a broad peak with fixed LSPR maximum (λ_max) at 560 ± 8 nm (Fig. 1A and D), attributed to a nanostructure with the prevalence of Au [2, 18]. Increasing p-CM amounts led to LSPR intensity increase and peak broadening; this is due to larger light scattering, classically related to multi-metal structure enlargements [19]. Coherently with the constant λ_max, the colloids retain the same color hue (purple).

RM, a conjugated ortho-diphenol, gives rise to a narrow absorption peak that, for increasing concentration, shows a blue shift with λ_max moving from 550 ± 6 nm to 510 ± 5 nm (Fig. 1B and E); this plasmonic profile is typical of Ag/Au alloys where the resonance of the two nanometals is merged [18–21]. The λ_max shift indicates an increase in the Ag-amount in the BiMNPs [18]. This BiMNP compositional metal evolution is confirmed by the shift of the colloid color from purple to orange.

CT, an ortho-diphenolic flavanol, displays two peaks that indicate the contemporary synthesis of AgNP (λ_max = 430 ± 3 nm) and AuNP (λ_max = 540 ± 4 nm) that resonate independently (Fig. 1C and E) [2, 18]. The AgNP occurs at higher CT amounts, and this triggers the colloid color switch from purple to ruby-red.

To better understand the BiMNPs formation route, the synthesis was attempted with reaction media containing only Ag⁺ (Fig. S1C) and Au³⁺ (Fig. S1D). No AgNPs were observed for the three PCs tested, while a slight AuNP signal was observed for p-CM and RM. This suggests that the BiMNP production is probably triggered by small AuNPs that work out as nucleation centers for Ag nanostructuration/synthesis [3, 15]. Eventually, to better attribute the observed LSPR profile to the different nanostructures, the Turkevich method (use of citrate as reduction agent) was employed to produce AgNPs, AuNPs, mixed populations of AgNPs/AuNPs, and Ag@Au core-shell NPs [2, 18]. The obtained plasmonic profiles (Fig. S2) correlate with those obtained with the PC, confirming their ability to drive different nanostructure generation according to the PC chemical structure.

The morphological and chemical nature of BiMNPs were further investigated by transmission electron microscopy (TEM), and X-ray photoelectron spectroscopy (XPS); Fig. 2A-C shows the TEM micrographs of BiMNPs synthesized with p-CM (A), RM (B), and CT (C) revealing a different nanostructuration.

The bi-metallic nature of the obtained nanoparticles is highlighted by the different electron densities of Au and Ag atoms, this results in a distinct contrast where the gold is darker than silver [15, 22]. As hypothesized by the LSPR spectra, the nanostructure arrangement obtained is related to the PC employed for the synthesis.

p-CM leads to spherical-shaped Ag/Au MNPs with heterogeneous Ag and Au arrangement (Fig. 2A); spherical MNPs are the most stable energetically, and their formation is reported for small molecules with low AoC (e.g., citrate, ascorbic acid, sodium borohydride, phenolic acids, etc.) [2, 23]. Indeed, p-CM is the simplest and smallest cinnamic acid with low antioxidant capacity [16]. BiMNPs synthesized with RM result in a structure often named 'nanosponge', where the Ag and Au are intimately co-arranged (Fig. 2B) [24]; this can be attributed to the RM conformational flexibility and terminal ortho-phenolic moieties which enables the formation of complex metal nanostructure [25]. On the other hand, CT is characterized by strong reducing capacity but higher structural rigidity, caused by the adjacent benzenic rings; this, as reported in Fig. 3C, leads to the formation of an AgNP/AuNP mixed population, which tends to be grouped in aggregates [23].

XPS analysis was performed to deepen the chemical state of the BiMNPs. Figure 2D-F reports the Ag3d and Au4f spectral region of BiMNPs synthesized with p-CM (D), RM (E), and CT (F). XPS high-resolution spectra of PCs-mediated BiMNPs confirm the copresence of Ag and Au, reporting the two spin-orbit coupling of Ag (3d_3/2 and 3d_5/2) and Au (4f_5/2 and 4f_7/2) [26, 27]. As control, XPS analysis was also performed on AgNP, AuNP, mixed populations of AgNP/AuNP, and Ag@Au core-shell NPs formed with citrate according to Turkevich method. Furthermore, monometallic AuNP and AgNP were formed using the three PCs studied when possible.

The observed binding energies (BE) are listed in Table S2; for all the BiMNPs formed with PC both Ag and Au are at the metal/zero valent state, and are characterized by a constant splitting doublet of 6.0 eV and 3.7 eV [26, 27]. Among the employed PC, RM and CT preserve the binding energies observed for monometallic AgNP and AuNP; whereas p-CM displays a behavior similar to Ag@Au core-shell NPs formed with the citrate, suggesting the formation of a similar nanostructure [26–28]. The Au/Ag ratio of the different BiNMPs was also calculated, using Ag3d_5/2 and Au4f_7/2 peak intensities (Table S3). As expected, CT demonstrated the higher Au/Ag ratio because of the co-presence of distinct monometallic NPs, followed by RM and p-CM; the lower Au/Ag observed for p-CM can be attributed to the Ag@Au core-shell arrangement. In brief, XPS analysis confirms the relationship between the PC-structure with the BiMNPs formation and nanostructure arrangement.

3.2 Plasmonic indexes definition

The reactivity of the different bimetallic synthesis mixtures: mix 1 (M1), mix 2 (M2), mix 3 (M3), and mix 4 (M4), characterized by different metal precursors ratios (section 2.4), were tested toward 9 PCs belonging to different chemical classes i.e. mono-phenols (p-HB, p-CM, NR), ortho-diphenols (CA, RM, CT), and trihydroxy-phenols (GA, EL, TN); the PCs structure and acronyms are reported in Fig. 3A, while classes and sub-classes are listed in Table S1.

Figure 3B shows the picture of the whole set of BiMNPs, while Fig. 3C reports the UV-Vis spectra obtained for M1; spectra for M2, M3, and M4 are reported in Fig. S3. The different reaction mixtures give rise to BiMNPs characterized by different colors and LSPR signatures. The metal precursor ratios also affected the bimetallic nanocolloidal arrangement. M1 gave rise to colloids for the whole set of tested PCs; M2 and M3 exhibited a higher reactivity towards PCs with high molecular weight (i.e., TN, CT, and NR); whereas M4 was more sensitive to ortho-phenolic structures (i.e., CT) with more resolved LSPR signatures compared to the other reaction mixtures.

According to LSPR signatures collected from the bimetallic colloids (Fig. 3C and Fig. S3), the maximum information concerning the BiMNP arrangements is given at wavelengths of 430, 510, 550, 600, and 650 nm; these have been already associated with Au- and Ag-based NPs characterized by different nanoarchitectures [18, 21]. To extract the maximum information to evaluate the PC profile pattern, improve the recognition ability, and amplify the LSPR variations, four plasmonic indexes (PI) were introduced.

The PIs λ₄₃₀, λ₅₅₀, and λ₆₅₀ are obtained by the maximum LSPR absorbance associated with a specific colloidal structure, divided by the absorbance at the relative baseline wavelength (see section 2.4). In particular, λ₄₃₀ is related to the formation of isolated silver nanoparticles (AgNP/AuNP mixed population); λ₅₅₀ is usually associated to bimetallic nanostructures with the prevalence of Au; λ₆₅₀ is characteristic of spherical-shaped BiMNPs with a heterogeneous Ag and Au arrangement. The PI λ₅₁₀ was obtained by the ratio of the 2 peaks at 510 nm and 550 nm; this evidences the presence of Ag in the BiMNP structure, in fact, PI λ₅₁₀ can be associated with bimetallic nanostructures where Ag and Au are intimately co-arranged (i.e., nanosponges).

3.3 Phenolic clustering in food using plasmonic Indexes

The ability of the plasmonic indexes selected to return information in food samples was challenged carrying the BiMNPs synthesis in 13 samples; in particular, ground and whole cloves (Cl1 and Cl2, respectively), coffee powders of different brands (Cf1, Cf2, and Cf3), roasted coffee beans’ peel (Cf4), green teas of various brands (Gt1, Gt2, and Gt3), dried mint leaves and mint infuse (Mn1 and Mn2, respectively), dried sage (Sg) and dried thyme (Ty) have been tested.

To have a quantitative idea of the PC patterns for each sample, HPLC–MS/MS targeted analysis according to a method already developed by our group was also run [29]. Table S4 reports the data obtained; a total of 31 different PCs were identified and quantified. Table S5 lists the PCs amount grouped for class (mono-phenols, ortho-diphenols, trihydroxy-phenols) and sub-classes (Benzoic acid, Ba; Cinnamic acid, Ca; Flavonoid, Fld; Benzoic/Cinnamic acid, Ba/Ca; Conjugated Ca; Conjugated Ba) according to their chemical structure. The samples present a heterogeneous composition characterized by a prevalent class of PCs; this is in agreement with the literature [30–33].

A classification of the food samples based on the PCs quantified via HPLC-MS/MS was attempted using the unsupervised multivariate hierarchical clustering (AHC) algorithm; the detailed procedure used is reported in section 2.7. The resulting heatmap of Fig. 4 shows that the AHC algorithm clustered the 13 food samples in four groups, named cluster-a, -b, -c, and -d.

The clusters' distance was computed using the Dunn index, and resulted equal to 1.22, proving a clear separation. In detail, cluster-a includes samples characterized by the prevalence of trihydroxy-phenols (i.e., Cl1 and Cl2) and cluster-b samples with a high relative abundance of chlorogenic acid (i.e., Cf samples); cluster-c is characterized by a remarkable amount of flavonoids with trihydroxy- (i.e., epigallocatechin and epigallocatechin gallate) and ortho-phenolic moieties (i.e., catechin). Samples with high content of ortho-diphenols, (particularly rosmarinic acid) are grouped in cluster-d (i.e., Mn1, Mn2, Sg, and Ty). Taking a look at the heatmap, both cluster-a and b are characterized by a lower total phenol content (TPC) with respect to cluster-c and cluster-d.

These sample extracts, at different dilutions (from 1:10 to 1:1000), were also used to produce BiMNPs in the proposed array format. Fig. S4, as an example, reports the plasmonic spectra obtained with the reaction mix M3 for samples Cl1, Mn1, Gt1, Cf1, and Sg1; Fig. S5 reports the pictures of the complete BiMNPs set. The colloidal suspensions, as expected, resulted in different colors that can be attributed to different BiMNP nanostructure arrangements, generated by the different PC patterns of the samples. The values of the four plasmonic indexes (PI: λ₄₃₀, λ₅₁₀, λ₅₅₀, and λ₆₅₀) proposed in section 2.2, extrapolated from the BiMNPs spectra of the samples were quite reproducible (RSD ≤ 4.4%; n = 3). In Fig. S6, it is possible to appreciate how PIs allow clear visualization of the prevalent PC classes of the samples.

The ability of the proposed PIs to identify the prevalent phenolic class in food samples was initially evaluated using Principal Component Analysis (PCA). Figure 5A and 5B report the score and loading plots for PC1, PC2, and PC3 respectively, obtained employing as dataset the PIs of pure phenolic compounds from the four reaction mixtures (M1, M2, M3, and M4); all data employed for statistical analysis are obtained from triplicates. The loadings represent the contribution of each PI to the principal components.

As shown in Fig. 5A, the PC1 axis highlighted the difference among the PIs (i.e., λ₄₃₀, λ₅₁₀, λ₅₅₀, and λ₆₅₀) representing 49.6% of the explained variance; PC2 and PC3 also give their contribution to the spatial separation of the PIs, expressing 22.2% and 17.8% of the variance, respectively; these data, demonstrate the different information brought by the selected plasmonic indexes that cover the 89.6% of the explained variance. PIs derived from the four reaction mixtures are grouped according to the wavelength, resulting spread in all the quadrants, indicating their ability to return different information.

In the score plot of Fig. 5B, which gives information on the phenolic compounds, low molecular weight phenolic structures with carboxylic acid functions (p-CM, p-HB, CF, and GA) are separated by the other phenolic compounds along PC1. Polyphenols with a dimeric or oligomeric structure (i.e., RM, EL, and TN) are well separated along PC2. Interestingly, despite their similar structures, NR and CT are separated, this is due to their different antioxidant capacity (AoC); indeed, NR is a low-reactive mono-phenol, and CT is an ortho-diphenol. From these data, it is clear that PIs give different outputs according to the reaction mix composition, depending on the structure and AoC of the single phenolic compound [5].

Figure S7A and S7B report loading and score plots obtained by the full set of PIs collected from food samples. In this case, despite the explained variance being higher (52.8% PC1, 28.9% PC2, and 8.1% PC3) and the PIs results spread along the quadrants, the majority of the samples are grouped regardless of their phenolic pattern not allowing a clear classification of the samples.

The classification ability of the PIs was then tested via partial least squares discriminant analysis (PLS-DA), to assess the indexes' ability to assign the samples to the four clusters (a, b, c, and d) defined with HPLC-MS/MS; a ‘leave one out’ procedure was used in cross-validation. The detailed PLS-DA procedure applied is described in section 2.7. PLS-DA is a descriptive modeling that emphasizes differences in datasets through a supervised algorithm enabling data classification [34]. To increase the performance and overcome the overfitting effect, PLS-DA was carried out by a k-fold (k = 10) cross-validation strategy [10, 35]; the classification performance of predictors conducted by permutation tests (100 cycles) is reported in Table S6 where the high value of the predictive parameter indicates the discriminant accuracy of the model (Q²Y ≥ 0.974) and the success of the classification.

The data of PLS-DA are appreciable in Fig. 6. From the 3 components obtained (total variance explained 88.1%) is clear that cluster-c is well-classified thanks to the opposite contribution of λ₅₅₀ and λ₆₅₀ in the 1st component; on the 2nd component cluster-a is discriminated by the inversely proportional contribution of λ₅₁₀ and λ₄₃₀, whereas cluster-b is well explained by λ₄₃₀. Correct classification of cluster-d is achieved by the synergic and positive contribution of all the PI explained by 3rd component.

Table S7 lists the PLS-DA data regarding the classification in fitting and cross-validation pointing out that the discriminant multivariate approach using PIs from the four reaction mixtures, taken one by one, did not accurately predict the food samples classification. This observation proves that to predict the representative phenolic classes of samples, consistent with clusters achieved by HPLC-MS/MS analysis, the synergic contribution of the PIs from the four reaction mixtures needs to be taken into account.

3.4 Smartphone-based calibration-free colorimetric score for antioxidant capacity evaluation

According to the plasmonic profile, BiMNPs return naked-eye visible colors. Using a simple smartphone camera, BiMNPs array color was used to evaluate the samples' antioxidant capacity (AoC). The rationale of this approach relies on the PC's ability to drive the formation of nanocolloidal ‘probes’ with different color-hue and hue-intensity (HI) according to the phenolic class and amount present in the sample, respectively, resulting in a colorimetric tag array.

RGB hue intensities were collected for each BiMNP colloids. To define an antioxidant capacity total score (AoC T.S.), the BiMNPs color was parametrized, dividing the different responses into four quartiles (Q) associated with a different score and colorimetric tag, coherently with the RGB hue collected: Q1 = score 1, gray-tag; Q2 = score 2.5, yellow-tag; Q3 = score 5, red-tag; Q4 = score 10, black-tag. The detailed quartiles definition and AoC T.S. extrapolation are described in section 2.8. and graphically resumed in Scheme 1. Fig. S8 reports the score grid for the full sets of PC standards (pure compounds) along with the extrapolated AoC T.S.; interestingly, the BiMNPs array profile and the relative AoC T.S. are different and peculiar for each PC (Fig. S5). The AoC T.S. was able to distinguish the different PCs according to their antioxidant capacity: TN > CT > NR > RM > GL ~ EL > CF > p-CM > p-HB. This PC antioxidant classification ability is in agreement with classical photometric assays [5].

The bimetallic colorimentric score was tested for food sample analysis; in this case, the score grid columns report different sample dilutions (D.F.). Figure 7 reveals the ability of the colorimetric score to differentiate the samples, and reveal the AoC; Fig. S5 shows the pictures from which the data have been collected.

Table S8 lists the AoC T.S. obtained for the 13 samples, in all the cases reproducible data were obtained (RSD ≤ 5.4, n = 3). Higher AoC T.S. was registered for green teas, followed by mints, ground cloves, coffee powders and dried sage. As expected, the samples with the lower score resulted the whole cloves, dried thyme, and roasted coffee peal. The food reactivity trend is coherent with the representative PCs and their concentration (Table S5). Indeed, green tea samples are the more reactive due to their phenolic pattern rich in ortho-diphenolic structure and flavonoids functionalized with ortho- and trihydroxylic groups, whereas ground clove is among the highest reactive sample due to the presence of the three-hydroxyphenols pattern and their content.

To further shed light on the information returned by the BiMNPs array, samples were analyzed with three consolidated photometric assays for total phenolic content (FC), antioxidant capacity (ABTS), and total flavonoids (AlCl₃) evaluation; the detailed procedures of the conventional methods used are given in section SI.2.6, while the data obtained are reported in Table S8. Sample classification is totally in agreement with the three assays, and appreciable correlations were obtained comparing the data with FC (r = 0.87), and ABTS (r = 0.82); interestingly, the AoC T.S. encloses information concerning both AoC and total phenol content. As expected, the bimetallic colorimetric array does not highlight a strong correlation with the AlCl₃-based assay (r = 0.61), since this method enables the assessment of flavonoids only.

Noteworthy, the bimetallic nanoparticle score is a calibration-free system able to return absolute scores concerning PC AoC, and to classify samples containing complex phenolic patterns without the need for external calibration and reference standards.

Herein, the production of plasmonic active colored bimetallic colloidal nanoparticles (BiMNPs) in an array format is proposed for the untargeted evaluation of polyphenol compounds' pattern and antioxidant capacity in food samples. The BiMNPs, generated by phenolic compounds (PC) in different reaction mixtures at different dilutions produce an array of absorption spectra and colors that are used for the classification. The rationale relies on the PC's ability to drive the formation of bimetallic silver/gold nanocolloids with different conformations; the peculiar plasmonic signatures were parametrized via the definition of plasmonic indexes (PI), while the nanocolloids color was used to define an antioxidant capacity total score (AoC T.S.) which only need a smartphone with a camera for the interpretation.

The ability of the BiMNPs-derived PIs to classify PCs in clusters defined with HPLC-MS/MS was tested toward 9 phenolic standards and 13 food samples; PLS-DA proved the capacity of the PI array to classify food samples according to the PC prevalent class. On the other hand, the colorimetric response acquired with a smartphone was able to return an AoC T.S. that allows to easily acquire information on the food polyphenols' antioxidant capacity; data obtained resulted in agreements with conventional photometric assays that need more time and external calibration.

Summing up, the BiMNPs array, according to the PCs’ amount and class, gives rise to nanocolloids with peculiar optical intensity and nano-conformation, which in turn return characteristic plasmonic and colorimetric tags, that enclose quali-quantitative information on the polyphenolic pattern of food samples. To the best of our knowledge, this is the first study where a nanocolloidal assay allows to obtain quali-quantitative information concerning food polyphenols exploiting simple experimental procedures for the formation of bimetallic nanoparticles.

Acknowledgments

This study received funding from the European Union - Next-GenerationEU - National Recovery and Resilience Plan (NRRP) – MISSION 4 COMPONENT 2, INVESTIMENT N. 1.1, CALL PRIN 2022 PNRR D.D. 1409 14-09-2022 – (Project No. P2022T3HFA_02; HYDEAL4safety) CUP C53D23007760001.

Ethics approval. Not applicable

Conflict of interest. The authors declare no competing interests.

Funding. Not applicable

Availability of data and material. Data will be made available on request.

Appendix

Supplementary information.

Supplementary materials and methods sections: SI.2.1 and 2.3 Reagents and apparatus, SI.2.5 BiMNPs characterization; SI.2.6 PCs evaluation with conventional analysis. Supplementary figures: Fig. S1. BiMNPs synthesis optimization; Fig. S2. Model BiMNPs investigation; Fig. S3. BiMNPs plasmonic profile comparison; Fig. S4. Plasmonic profile of BiMNPs synthesized with samples extract; Fig. S5. Pictures of BiMNPs synthesized with samples extract; Fig. S6. Plasminic indexes comparison; Fig. S7. Principal Component Analysis of samples-synthesized BiMNPs; Fig. S8. BiMNPs colorimetric array results obtained for the different PCs. Supplementary tables: Tab. S1. PCs acronyms; Tab. S2. Ag3d and Au4f binding energies from XPS spectra; Tab. S3. Au/Ag ratio calculated from XPS spectra; Tab. S4. PCs determined by HPLC-MS/MS in food samples; Tab. S5. Relative concentration for the phenolic chemical class; Tab. S6. PLS-DA classification/regression performance; Tab. S7. PLS-DA classification in fitting and cross-validation; Tab. S8. Total phenolic content (FC), antioxidant capacity (ABTS), total flavonoids (AlCl₃), and antioxidant capacity total score (AoC T.S.) obtained for the food samples.

Scroccarello A, Della Pelle F, Del Carlo M, Compagnone D (2023) Optical plasmonic sensing based on nanomaterials integrated in solid supports. A critical review. Anal Chim Acta 1237:340594. https://doi.org/10.1016/j.aca.2022.340594
Mart R, Ramos CR, Regulacio MD (2021) Controllable Synthesis of Bimetallic Nanostructures Using Biogenic Reagents : A Green Perspective. https://doi.org/10.1021/acsomega.1c00692
Loiseau A, Zhang L, Hu D, et al (2019) Core-Shell Gold/Silver Nanoparticles for Localized Surface Plasmon Resonance-Based Naked-Eye Toxin Biosensing. ACS Appl Mater Interfaces 11:46462–46471. https://doi.org/10.1021/acsami.9b14980
Peng B, Zhou J, Xu J, et al (2019) A smartphone-based colorimetry after dispersive liquid–liquid microextraction for rapid quantification of calcium in water and food samples. Microchem J 149:104072. https://doi.org/10.1016/j.microc.2019.104072
Della Pelle F, Scroccarello A, Sergi M, et al (2018) Simple and rapid silver nanoparticles based antioxidant capacity assays : Reactivity study for phenolic compounds. Food Chem 256:342–349. https://doi.org/10.1016/j.foodchem.2018.02.141
Bordbar MM, Hemmateenejad B (2018) An optoelectronic tongue based on an array of gold and silver nanoparticles for analysis of natural , synthetic and biological antioxidants. Microchim Acta 185:493
Zhang Q, Li N, Goebl J, et al (2011) A systematic study of the synthesis of silver nanoplates: Is citrate a “magic” reagent? J Am Chem Soc 133:18931–18939. https://doi.org/10.1021/ja2080345
Li S, Li K, Li X, Chen Z (2019) A colorimetric electronic tongue for rapid discrimination of antioxidants based on the oxidation etching of nano-triangular silver by metal ions. ACS Appl Mater Interfaces 11:37371–37378. https://doi.org/10.1021/acsami.9b14522
Li L, Li S, Yu X, Chen Z (2021) Visual detection of multiple antioxidants based on three chloroauric acid/Au-Ag nanocubes. Microchim Acta 188:. https://doi.org/10.1007/s00604-021-04774-5
Team RC (2015) R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing; 2014. R Foundation for Statistical Computing. R Found Stat Comput Vienna, Austria 2:2019
Brock G, Pihur V, Datta S, Datta S (2008) clValid: An R Package for Cluster Validation Guy. J Stat Softw 25:371–372. https://doi.org/10.1016/0038-1098(77)91248-0
Kucheryavskiy S (2020) mdatools – R package for chemometrics. Chemom Intell Lab Syst 198:103937. https://doi.org/10.1016/j.chemolab.2020.103937
Cacciatore S, Tenori L, Luchinat C, et al (2017) KODAMA: An R package for knowledge discovery and data mining. Bioinformatics 33:621–623. https://doi.org/10.1093/bioinformatics/btw705
Grzesik M, Naparło K, Bartosz G, Sadowska-Bartosz I (2018) Antioxidant properties of catechins: Comparison with other antioxidants. Food Chem 241:480–492. https://doi.org/10.1016/j.foodchem.2017.08.117
Wang Y, Zeng Y, Fu W, et al (2018) Seed-mediated growth of Au@Ag core-shell nanorods for the detection of ellagic acid in whitening cosmetics. Anal Chim Acta 1002:97–104. https://doi.org/10.1016/j.aca.2017.11.067
Tsao R (2010) Chemistry and biochemistry of dietary polyphenols. Nutrients 2:1231–1246. https://doi.org/10.3390/nu2121231
Jeon H Bin, Tsalu PV, Ha JW (2019) Shape Effect on the Refractive Index Sensitivity at Localized Surface Plasmon Resonance Inflection Points of Single Gold Nanocubes with Vertices. Sci Rep 9:1–8. https://doi.org/10.1038/s41598-019-50032-3
Ristig S, Prymak O, Loza K, et al (2015) Nanostructure of wet-chemically prepared, polymer-stabilized silver-gold nanoalloys (6 nm) over the entire composition range. J Mater Chem B 3:4654–4662. https://doi.org/10.1039/c5tb00644a
R. Daniel J, McCarthy LA, Ringe E, Boudreau D (2019) Enhanced control of plasmonic properties of silver-gold hollow nanoparticles via a reduction-assisted galvanic replacement approach. RSC Adv 9:389–396. https://doi.org/10.1039/C8RA09364D
Kalantari K, Fahimi-Kashani N, Hormozi-Nezhada MR (2022) Development of a colorimetric sensor array based on monometallic and bimetallic nanoparticles for discrimination of triazole fungicides. Anal Bioanal Chem 414:5297–5308. https://doi.org/10.1007/s00216-021-03272-0
Al-Azawi MA, Bidin N, Bououdina M, Mohammad SM (2016) Preparation of gold and gold-silver alloy nanoparticles for enhancement of plasmonic dye-sensitized solar cells performance. Sol Energy 126:93–104. https://doi.org/10.1016/j.solener.2015.12.043
Meena Kumari M, Jacob J, Philip D (2015) Green synthesis and applications of Au-Ag bimetallic nanoparticles. Spectrochim Acta - Part A Mol Biomol Spectrosc 137:185–192. https://doi.org/10.1016/j.saa.2014.08.079
Scroccarello A, Molina-Hernández B, Della Pelle F, et al (2021) Effect of phenolic compounds-capped AgNPs on growth inhibition of Aspergillus niger. Colloids Surfaces B Biointerfaces 199:. https://doi.org/10.1016/j.colsurfb.2020.111533
Ismail WZW, Dawes JM (2022) Synthesis and Characterization of Silver-Gold Bimetallic Nanoparticles for Random Lasing. Nanomaterials 12:1–11. https://doi.org/10.3390/nano12040607
Stasiłowicz-Krzemień A, Rosiak N, Płazińska A, et al (2022) Cyclodextrin Derivatives as Promising Solubilizers to Enhance the Biological Activity of Rosmarinic Acid. Pharmaceutics 14:. https://doi.org/10.3390/pharmaceutics14102098
Atar N, Eren T, Demirdögen B, et al (2015) Silver, gold, and silver@gold nanoparticle-anchored l-cysteine-functionalized reduced graphene oxide as electrocatalyst for methanol oxidation. Ionics (Kiel) 21:2285–2293. https://doi.org/10.1007/s11581-015-1395-1
Murugavelu M, Karthikeyan B (2017) Synthesis, characterization of Ag-Au core-shell bimetal nanoparticles and its application for electrocatalytic oxidation/sensing of L-methionine. Mater Sci Eng C 70:656–664. https://doi.org/10.1016/j.msec.2016.09.046
Gopalakrishnan R, Loganathan B, Raghu K (2015) Green synthesis of Au-Ag bimetallic nanocomposites using Silybum marianum seed extract and their application as a catalyst. RSC Adv 5:31691–31699. https://doi.org/10.1039/c5ra03571f
Oliva E, Viteritti E, Fanti F, et al (2021) Targeted and semi-untargeted determination of phenolic compounds in plant matrices by high performance liquid chromatography-tandem mass spectrometry. J Chromatogr A 1651:462315. https://doi.org/10.1016/j.chroma.2021.462315
Ali A, Bashmil YM, Cottrell JJ, et al (2021) Lc-ms/ms-qtof screening and identification of phenolic compounds from australian grown herbs and their antioxidant potential. Antioxidants 10:. https://doi.org/10.3390/antiox10111770
Chacko SM, Thambi PT, Kuttan R, Nishigaki I (2010) Beneficial effects of green tea: A literature review. Chin Med 5:1–9. https://doi.org/10.1186/1749-8546-5-13
El-Maati MFA, Mahgoub SA, Labib SM, et al (2016) Phenolic extracts of clove (Syzygium aromaticum) with novel antioxidant and antibacterial activities. Eur J Integr Med 8:494–504. https://doi.org/10.1016/j.eujim.2016.02.006
Król K, Gantner M, Tatarak A, Hallmann E (2020) The content of polyphenols in coffee beans as roasting, origin and storage effect. Eur Food Res Technol 246:33–39. https://doi.org/10.1007/s00217-019-03388-9
Lee LC, Liong CY, Jemain AA (2018) Partial least squares-discriminant analysis (PLS-DA) for classification of high-dimensional (HD) data: A review of contemporary practice strategies and knowledge gaps. Analyst 143:3526–3539. https://doi.org/10.1039/c8an00599k
Westerhuis JA, Hoefsloot HCJ, Smit S, et al (2008) Assessment of PLSDA cross validation. Metabolomics 4:81–89. https://doi.org/10.1007/s11306-007-0099-6

Scheme 1 is available in the Supplementary Files section.

No competing interests reported.

SCHEME1.jpg
Scheme 1. AoC T.S. scores and extrapolation. (A) Colorimetric quartiles: RGB% ranges, scores values and tags. (B) Example of the AoC T.S. extrapolation via BiMNPs array score grid for a generic sample; P.S. = partial score value, T.S. = total score value.
20240801BiMNPsSI.pdf
Appendix Supplementary information. Supplementary materials and methods sections: SI.2.1 and 2.3 Reagents and apparatus, SI.2.5 BiMNPs characterization; SI.2.6 PCs evaluation with conventional analysis. Supplementary figures: Fig. S1. BiMNPs synthesis optimization; Fig. S2. Model BiMNPs investigation; Fig. S3. BiMNPs plasmonic profile comparison; Fig. S4. Plasmonic profile of BiMNPs synthesized with samples extract; Fig. S5. Pictures of BiMNPs synthesized with samples extract; Fig. S6. Plasminic indexes comparison; Fig. S7. Principal Component Analysis of samples-synthesized BiMNPs; Fig. S8. BiMNPs colorimetric array results obtained for the different PCs. Supplementary tables: Tab. S1. PCs acronyms; Tab. S2. Ag3d and Au4f binding energies from XPS spectra; Tab. S3. Au/Ag ratio calculated from XPS spectra; Tab. S4. PCs determined by HPLC-MS/MS in food samples; Tab. S5. Relative concentration for the phenolic chemical class; Tab. S6. PLS-DA classification/regression performance; Tab. S7. PLS-DA classification in fitting and cross-validation; Tab. S8. Total phenolic content (FC), antioxidant capacity (ABTS), total flavonoids (AlCl₃), and antioxidant capacity total score (AoC T.S.) obtained for the food samples.

Download PDF

Editorial decision: Revision requested
28 Aug, 2024
Reviews received at journal
28 Aug, 2024
Reviewers agreed at journal
16 Aug, 2024
Reviews received at journal
14 Aug, 2024
Reviewers agreed at journal
13 Aug, 2024
Reviewers invited by journal
10 Aug, 2024
Editor assigned by journal
02 Aug, 2024
Submission checks completed at journal
02 Aug, 2024
First submitted to journal
01 Aug, 2024

You are reading this latest preprint version

Bimetallic nanocolloidal plasmonic array for polyphenols characterization and calibration-free antioxidant capacity evaluation

Status:

Version 1

Abstract

Figures

1. Introduction

2. Materials and methods

2.1. Reagents and stock solutions

2.2. Phenolic compounds and samples

2.3. Apparatus

2.4 Bimetallic nanocolloids synthesis and plasmonic indexes

2.5. Bimetallic nanocolloids physico-chemical characterization

2.6. Phenolic compound analysis via conventional methods

2.7. Phenolic compounds clustering and plasmonic indexes multivariate analysis

2.8. Smartphone-based antioxidant capacity score via colorimetric bimetallic array

3. Results and discussion

3.1. Bimetallic nanocolloids optical and morpho-chemical features

3.2 Plasmonic indexes definition

3.3 Phenolic clustering in food using plasmonic Indexes

3.4 Smartphone-based calibration-free colorimetric score for antioxidant capacity evaluation

4. Conclusions

Declarations

References

Scheme 1

Additional Declarations

Supplementary Files

Status:

Version 1