Validation of a New HPLC-DAD Method to Quantify 3-Deoxyanthocyanidins Based on Carajurin in Medicinal Plants and for Chemical Ecology Studies

Anthocyanidins are found in nature mainly as pelargonidin, cyanidin, peonidin, delphinidin, petunidin, and malvidin derivatives. These compounds are found free or as glycoside derivatives which are responsible for the color (red, blue, and violet) of some foods and are responsible for attracting seed dispersers. They are grouped into 3-hydroxyanthocyanidins, 3-deoxyanthocyanidins (3D-anth), and O-methylated anthocyanidins. A new method was developed and validated to quantify 3D-anth in plant-rich extracts. To test the new method, Arrabidaea chica Verlot was selected as it is widely used in folk medicine, and it is rich in 3D-anth. The new method was developed by HPLC-DAD and expressed 3D-anth as carajurin content. Carajurin was chosen as the reference standard due to its role as a biological marker for the antileishmanial activity for A. chica. The selected method used a silica-based phenyl column, a mobile phase composed of potassium dihydrogen phosphate buffer, acetonitrile, and methanol, in a gradient elution mode and detection at 480 nm. The method reliability was confirmed by verifying selectivity, linearity, precision, recovery, and robustness. This method contributes to quality control and development of a possible active pharmaceutical ingredient from A. chica as well as it can be used to evaluate 3D-anth in plant extracts with chemical ecology interest.


Introduction
Anthocyanidins feature a flavilium cation in their structure. Six of 19 different types of anthocyanidins (pelargonidin, cyanidin, peonidin, delphinidin, petunidin, and malvidin) prevail in nature ( Figure 1) [1]. These compounds are found in glycoside forms (anthocyanins) and are responsible for the color of some foods, especially fruits such as cranberry, blueberry, grape, sweet cherry, strawberry, marionberry, blackberry, raspberry, apple, peach, nectarine, and plum [2]. In plants, anthocyanidins offer protection from UV, and they are active against microorganisms, including virus, bacteria, and fungus. Besides, anthocyanidins may attract seed dispersers, an important ecological function. They are grouped into 3-hydroxyanthocyanidins, 3-deoxyanthocyanidins (3D-anth), and O-methylated anthocyanidins [3,4]. The 3D-anth differs from the other by the absence of a hydroxyl at position 3 of the C-ring. Considered the anthocyanins chemical precursors, they are found in flowers and fruits and responsible for their impressive blue and purple colors. Occurring in a limited number of plant species, including the medicinal Arrabidaea chica (Humb. & Bonpl.) absence of a hydroxyl at position 3 of the C-ring. Considered the anthocyanins chemical precursors, they are found in flowers and fruits and responsible for their impressive blue and purple colors. Occurring in a limited number of plant species, including the medicinal Arrabidaea chica (Humb. & Bonpl.) B. Verl., they show a redder color than the corresponding anthocyanins [5]. To date, the literature described the extraction, isolation, and chemical characterization of four 3D-anth from A. chica: 6,7,3′,4′-tetrahydroxy-5methoxy-flavylium (3′-hydroxy-carajurone) (1), 6,7,4′-trihydroxy-5-methoxy-flavylium (carajurone) (2), 6,7,3′-trihydroxy-5,4′-methoxy-flavylium (3′-hydroxy-carajurin) (3) and 6,7-dihydroxy-5,4′-dimethoxy-flavylium (carajurin) (4) [6,7]. Their chemical structures are shown in Figure 1. Despite having the same biosynthetic origin as other flavonoids, anthocyanidins have specific structural characteristics. Due to a continuous delocalization of the π electrons along the A, B and C rings of their backbone, the absorption band of anthocyanins occurs in the visible range, varying between 480 and 540 nm. In the specific case of 3D-anth, the highest absorption occurs at 480 nm [8][9][10]. From these data, a first screening was performed in a full-scan experiment (200-600 nm). As results, chromatograms at 480 nm show the maximum wavelength (λmax) for the 3D-anth, confirming the literature data; therefore, this wavelength was adopted in the monitoring of these class of substances.
Arrabidaea chica was chosen for these experiments because it is in the National List of Medicinal Plants of Interest (Renisus) to the Unified Health System (SUS) by the Brazilian Ministry of Health (MH) in 2009. This list includes 71 plant species being used in treatments or whose pharmacological activities have been proven in studies. According to the MH, these species have the potential to reach production and benefit the SUS, including A. chica [11]. Native to the Amazon region, A. chica, popularly named crajiru and pariri, belongs to the Bignoniaceae family. It is also found in American and African tropical regions [6,12]. Its leaves are traditionally used as teas to treat skin inflammation, gastrointestinal disorders, mycoses, wounds, ulcers, and other ailments [13][14][15].
Carajurin is the most studied 3D-anth from A. chica. Its isolation and chemical characterization are reported by several research groups [6,7,16]. Recently, carajurin was proposed as biological marker for the antileishmanial activity of hydroalcoholic extracts of A. chica, as activity levels against the promastigote forms of Leishmania amazonensis correlate with the carajurin content [16]. Subsequently, a bioguided study confirmed, when comparing activities of the crude hydroalcoholic extract from A. chica, an anthocyanidins-rich fraction, and its isolated anthocyanidins, that carajurin has the highest activity of all examined options [17]. Then, a set of experimental evidence and Despite having the same biosynthetic origin as other flavonoids, anthocyanidins have specific structural characteristics. Due to a continuous delocalization of the π electrons along the A, B and C rings of their backbone, the absorption band of anthocyanins occurs in the visible range, varying between 480 and 540 nm. In the specific case of 3D-anth, the highest absorption occurs at 480 nm [8][9][10]. From these data, a first screening was performed in a full-scan experiment (200-600 nm). As results, chromatograms at 480 nm show the maximum wavelength (λ max ) for the 3D-anth, confirming the literature data; therefore, this wavelength was adopted in the monitoring of these class of substances.
Arrabidaea chica was chosen for these experiments because it is in the National List of Medicinal Plants of Interest (Renisus) to the Unified Health System (SUS) by the Brazilian Ministry of Health (MH) in 2009. This list includes 71 plant species being used in treatments or whose pharmacological activities have been proven in studies. According to the MH, these species have the potential to reach production and benefit the SUS, including A. chica [11]. Native to the Amazon region, A. chica, popularly named crajiru and pariri, belongs to the Bignoniaceae family. It is also found in American and African tropical regions [6,12]. Its leaves are traditionally used as teas to treat skin inflammation, gastrointestinal disorders, mycoses, wounds, ulcers, and other ailments [13][14][15].
Carajurin is the most studied 3D-anth from A. chica. Its isolation and chemical characterization are reported by several research groups [6,7,16]. Recently, carajurin was proposed as biological marker for the antileishmanial activity of hydroalcoholic extracts of A. chica, as activity levels against the promastigote forms of Leishmania amazonensis correlate with the carajurin content [16]. Subsequently, a bioguided study confirmed, when comparing activities of the crude hydroalcoholic extract from A. chica, an anthocyanidins-rich fraction, and its isolated anthocyanidins, that carajurin has the highest activity of all examined options [17]. Then, a set of experimental evidence and literature reports consolidated the proposition of carajurin as a biological marker of A. chica for antileishmanial activity [18]. Lately, it showed lethal effect against promastigote forms of L. amazonensis, inducing ultrastructural changes, mitochondrial membrane potential decrease, and increased ROS production, which together induced cell death by late apoptosis [19].
Carajurin might be exclusive to the genus Arrabidaea and possibly limited to the species A. chica due to lack of mention of its isolation from other plant species [16,20]. But with new and fast validated methods for analysis of 3D-anth, other plant material can be monitored in a medicinal and chemical ecological point of views. Besides, despite long use and its Renisus-inclusion, no official compendium still regulates the quality control of A. chica. Quality is challenging when handling plant species, especially considering industry's aims of obtaining safe and effective pharmacological products. The World Health Organization (WHO) recommends methods to assess the quality of medicinal plant materials [21]. The complexity of extracts' chemical composition and their potential seasonal and regional variation requires plant materials' standardization. The WHO accepts chromatographic profile analyses for identifying and evaluating the quality of medicinal plants and the variability of the chemical components of extracts [22,23]. The standardization of extracts is essential for herbal medicines production, enabling the administration of known concentrations of compounds and better pharmacological responses [24]. This is a key step for their use as active pharmaceutical ingredients (API).
Considering chemical ecology studies, analytical methods capable of performing fast and robust analyzes can contribute to studies, for example, of substances that are responsible for the colors of the plants, including anthocyanidins. These studies are extremely important to verify if a seed disperser is also attracted by the color, consequently, by the anthocyanidin and anthocyanin content [3,4].
To date, only three papers show analytical methods by High-Performance Liquid Chromatography (HPLC) for quantification of anthocyanidins from A. chica [6,25,26]. However, they lack reproducibility, good resolution for the carajurin peak, and require long analysis times (30 to 45 min). None of them presented information about development and/or validation, i.e., they do not guarantee reliability.

Method Development
The analytical conditions tested ( Table 1) followed HPLC's method development procedures [27][28][29]. The first attempts to develop a practical method for quantification of 3D-anth from A. chica used different proportions of acidified water (H 2 O)/acetonitrile (ACN)/methanol (MeOH) (v/v/v) in isocratic conditions. However, none of the five different tested proportions provided visualization of peaks for carajurin and other anthocyanidins present in the extract. Thus, the same solvent mixture was tested with a gradient elution, and visualized 3D-anth's peaks, but being only registered at the end of each chromatogram. This led to longer analysis time (from 12 to 18 min). Aiming to improve resolution, asymmetry, and width of the peak base, further analytical conditions were changed: new mixtures of the eluents; chromatographic columns; flow of the mobile phase; and gradient programs; totaling 45 attempts. For each small change resulting in improvement of carajurin peak, the conditions were maintained; if not, the previous condition was returned, and tested a new condition, until obtaining the best analytical conditions ( Table 2). Figure 2 shows the resulting chromatogram of carajurin in the new method. The carryover test was monitored intra-day and inter-day by blank injections. A small peak was detected but its area in mAU was closer to limit of quantification (LDQ) ( Figure 3A).

Method Validation
After establishing the best chromatographic parameters, we proceeded with the analytical validation. The newly developed analytical method (Table 2) was validated for selectivity, linearity, intra-day and inter-day precision, accuracy, detection and quantification limits, recovery, and robustness [30,31].

Selectivity
Aiming to verify the absence of interferences in the detection window of the carajurin peak, the sample diluent (ACN:MeOH, 75:25, v/v) and an extract containing carajurin were evaluated. The chromatogram of the diluent ( Figure 3A) shows no peaks, while in the one for the A. chica hydroalcoholic extract (ACHE) (Figure 3B), there is no interference close to the Rt of carajurin, which demonstrates the selectivity of the method.

Linearity
Linearity was evaluated by analytical curves from carajurin diluted solutions (1; 2; 3; 4; 5, and 6 µg/mL). Analyses were in triplicate on three different days. No outliners were registered, and the average linear correlation coefficient (r 2 ) was 0.9970, higher than the minimum value (0.99) described by the adopted references [30,31], demonstrating the method's linearity. The linear regression equation used for quantification of samples was ABS = (104,771.3 ± 1873) × Concentration − (58,184 ± 9974). The points distribution in the analytical curve showed to be homoscedastic.

Intra-Day and Inter-Day Precision
In the evaluation of precision, three concentrations of carajurin standard (1, 3, and 6 µg/mL) were used in triplicate at three different times on the same day (intra-day precision) and three different days (inter-day precision). The triplicate means, the standard deviation (SD), and the relative standard deviation (RSD) of analysis were calculated (Tables 3 and 4). No significant differences were observed in results (p > 0.05). Additionally, considering that the variation values were lower than the established limits of RSD (<15%) [30,31], the method is precise.

Detection and Quantification Limits
The limits of detection (LOD) and quantification (LOQ) were evaluated through the relation between signal and noise from the baseline in the chromatogram for successive dilutions of the carajurin stock solution. According to the reference guides, the ideal signalto-noise ratios are 1:3 for LOD and 1:10 for LOQ. Considering these parameters, LOD was determined at 125 ng/mL and LOQ at 200 ng/mL (LOQ), therefore, compatible with a DAD-UV-Vis detector.

Accuracy
The accuracy of the method was evaluated as percentage comparison between the theoretical concentrations (analytical curve points) and the ones experimentally obtained. As the variations found are within the limits allowed by the validation standards (85-115%), the method can be considered accurate (Table 5).  4 ∆ total (%) = total difference in percentage from theorical concentration. Experimental concentration was calculated from the analytical curve formula described in Linearity.

Robustness
Four parameters and their variations were separately analyzed to evaluate robustness, meaning the entire method remained following nominal conditions, except for the variation of the parameter currently evaluated. This influence was evaluated at the lowest and highest concentration of the analytical curve. Acceptable variation is equal or bellow RSD of 15%. The highest variation was just over 2% and the method can be considered robust (Table 6).

Recovery
This parameter was evaluated using carajurin-spiked samples at the highest and lowest concentrations (1 and 6 µg/mL) and the percentage comparison between the standard and the enriched solutions. Table 7 shows the results. Recovery rates were within the specification range determined by the legal references used (85-115%).

Comparison of the Newly Developed and Validated Method with Those from Literature
Some HPLC methodologies address analyses of A. chica extracts [32][33][34][35][36][37][38]. However, few studies specify a methodology for quantification of anthocyanidins in this species (Table 8). Prior to developing our methodology, several attempts proved unsuccessful in reproducing them. Additionally, none of these studies described the development and validation of their methodologies. For example, those methods developed by Paula et al. [25,26], followed by Devia et al.'s methodology [6], differ from ours in column dimensions, flow rate, and total analysis time. Taffarello et al.'s HPLC-DAD method [39] lacked a complete specification of the elution gradient, which precludes the determination of the total analysis time, thus being unreproducible. The method presented herein was compared to those from the literature. It shows improvements concerning the peak width of the carajurin (approximately 1.0 min in our method vs. about 3.0 min in Devia et al. [6]), and total analysis time (18.0 min in this study vs. 45.0 min in Paula et al. [25]). These enhancements mean better chromatographic resolution and significant time saved, enabling more analyses per time and, consequently, lower mobile phase consumption. Other parameters were calculated to carajurin standard, such as peak asymmetry (0.9), peak tailing factor (1.1) and UV spectrum purity (98%).

Quantification of 3-Deoxyanthocyanidins (3D-Anth) Expressed as Carajurin
We quantified the 3D-anth in ACHE using the new developed and validated analytical method. The objective was to test the quantification efficacy of the new method in a more complex matrix, in addition to the tests already carried out during the development and validation stages. Figure 3B shows the obtained chromatogram of ACHE. Four 3D-anth were assigned (1)(2)(3)(4), of which we identified carajurin (3) Table 9. A previous comparison between four morphotypes of A. chica in two seasonal collections [16], quantifying carajurin in all extracts, took 90 min, a much longer analysis time. Further seasonal analyses based on the herein presented method will be more cost-effective.
Once adjusted, this method may be used to quantify 3D-anth in other plant materials or 3D-anth-rich extracts, considering its relevant gain in resolution and analysis time.

Plant Material Collection and Identification
Leaves from A. chica were cultivated and collected at Farmanguinhos/Fiocruz Atlantic Forest Campus, municipality of Rio de Janeiro, State of Rio de Janeiro, Brazil (S22.9406 • W43.4046 • ). Plant material was identified by Dr. Marcus Felipe Oliveira da Silva at the Botanical Collection of Medicinal Plants (CBPM) of Farmanguinhos/Fiocruz, where a voucher specimen is deposited and registered under the number CBPM668. The plant material was compared with those deposited at the Herbarium of the Botanical Garden of Rio de Janeiro (JBRJ) and CBPM. All studies with A. chica were registered at the Genetic Heritage Management Council under the code A670412.

Extract Preparation and Carajurin Isolation
One gram of ground dried leaves (knife mill after oven at 60 • C with forced air circulation) was extracted by maceration in ultrasonic bath (Quimis, Diadema, Brazil) at room temperature in 10 mL ethanol:water (7:3, v/v) for 30 min. The obtained ACHE was filtered in a 0.45 µm PTFE filter (Merck Millipore, Darmstadt, Germany) before analyzed in a High-Performance Liquid Chromatograph coupled to Diode-Array Detector (HPLC-DAD). Carajurin was isolated and characterized [16]. The determined purity was 98% by HPLC-DAD [16].
In addition, we examined mobile phase flow, isocratic or gradient elution mode, and total time analysis, in this validation. Keeping oven temperature at 50 • C, analyses were performed in triplicate (20 µL sample injection), and 3D-anth detected at 480 nm.

Method Validation
We followed the guidance for validation from the resolution RDC 166/2017 of the Brazilian Health Surveillance Agency-Anvisa [30] and DOQ-CGCRE-008 of the Brazilian National Institute of Metrology, Quality and Technology (Inmetro) [31]. Analytical parameters evaluated were selectivity, linearity, intra-day and inter-day precision, accuracy, detection and quantification limits, robustness, and recovery. Isolated carajurin was used as the reference standard. Analytical curves were obtained from carajurin standard solutions-stock solution (200 µg/mL) in acetonitrile:methanol (75:25, v/v); and at 1, 2, 3, 4, 5, and 6 µg/mL (final concentrations). All determinations were done with a 20 µL injection volume and detection at 480 nm. Selectivity was obtained from the analysis of sample diluent (acetonitrile:methanol, 75:25, v/v), carajurin solution, and A. chica carajurincontaining extract samples. Linearity was achieved from three calibration curves obtained on three different days. The evaluation of precision was with three concentrations (1, 3, and 6 µg/mL) in triplicate at three different times on the same day (intra-day precision) and in three different days (inter-day precision). Accuracy was evaluated as percentage comparison between theoretical concentrations (analytical curve points) and the experimentally obtained. Limits of detection (LOD) and quantification (LOQ) were established from serial dilutions and observation of the ratio signal/noise (s/n) (LOD: s/n ≥ 3; LOQ: s/n ≥ 10). Robustness tests were carried out based on the analysis of the influence of four analytical parameters (mobile phase flow rate; oven temperature; brand of acetonitrile; and detection wavelength) and their respective variations in two different concentrations (1 and 6 µg/mL). Recovery was obtained from carajurin-spiked samples at the highest and lowest concentrations (6 and 1 µg/mL, respectively), followed by determining the concentrations added.

Preparation of Extract Samples
A total of 1000 µL of acetonitrile:methanol (both HPLC grade, Tedia) (75:25, v/v) was added to 10 mg of each extract, previously weighed in a 4 mL vial. The vial was sealed and brought to ultrasonic bath for 10 min with occasional swirling. The sample was vortexed to mix thoroughly and filtered (0.45 µm PTFE filter; Merck, Darmstadt, Germany) before further analyses.

Statistical Analysis
Results are noted as absorbance mean and standard deviation. Tukey's test (ANOVA) was used to compare intra-day and inter-day precision (p < 0.05) using STATISTICA software version 10 (StartSoft Inc., Tulsa, OK, USA).

Conclusions
This paper presents a newly developed and validated method using HPLC-DAD to quantify 3D-anth expressed as carajurin. It showed, as a major advantage, a better peak definition of carajurin given by the peak asymmetry, tailing factor and peak width of this compound. In addition, the present method proves to be faster and more economical than those already existing in the literature, with a shorter total analysis time and in addition to the use of a silica-based phenyl column. Its validation demonstrated reliability by selectivity, linearity, precision, accuracy, recovery, and robustness, which demonstrates that the method is reproducible. Thus, the results will contribute to setting parameters for the quality control of medicinal plants rich in 3D-anth, such as A. chica, as well as anthocyanidins-rich extracts, to obtain active pharmaceutical ingredients or for chemical ecology studies.