Characterization and Quantification of the Compounds of the Ethanolic Extract from Caesalpinia ferrea Stem Bark and Evaluation of Their Mutagenic Activity

Caesalpinia ferrea Martius has traditionally been used in Brazil for many medicinal purposes, such as the treatment of bronchitis, diabetes and wounds. Despite its use as a medicinal plant, there is still no data regarding the genotoxic effect of the stem bark. This present work aims to assess the qualitative and quantitative profiles of the ethanolic extract from the stem bark of C. ferrea and to evaluate its mutagenic activity, using a Salmonella/microsome assay for this species. As a result, a total of twenty compounds were identified by Flow Injection Analysis Electrospray Ionization Ion Trap Mass Spectrometry (FIA-ESI-IT-MS/MSn) in the ethanolic extract from the stem bark of C. ferrea. Hydrolyzable tannins predominated, principally gallic acid derivatives. The HPLC-DAD method was developed for rapid quantification of six gallic acid compounds and ellagic acid derivatives. C. ferrea is widely used in Brazil, and the absence of any mutagenic effect in the Salmonella/microsome assay is important for pharmacological purposes and the safe use of this plant.


Identification of Constituents by FIA-ESI-IT-MS/MS n
In total, twenty-six compounds were identified by FIA-ESI-IT-MS/MS n (Figure 1) in the ethanolic extract of the stem bark of C. ferrea. Data concerning the identification of the peaks are shown in Table 1, in which we report the retention time, UV-Vis absorptions and electrospray ionization mass spectrometry in negative ion mode for all of the compounds detected. Since this is the first experiment for the chemical characterization of most of the constituents in the ethanolic extract from the stem bark of C. ferrea, it was necessary to characterize them completely.   [42] The UV spectra and fragmentations in the second order spectra (Table 1) show that the ethanolic extract consists mainly of hydrolyzable tannins, when compared with the literature data for these compounds [19,23,39,43].
Gallic acid and several of its derivatives were identified in the extract when compared to the Rt, UV and MS spectra. Compound 2 was assigned to gallic acid. The deprotonated molecule ion at m/z 169 confirmed the presence of this compound [23,24]. The deprotonated molecule ion at m/z 197 was identified with an Rt of 29.01 min. Second order spectra of this ion led to the formation of the ion product at m/z 169, which is related to the loss of CH2 = CH2 [M − 28 − H] − and the ion fragment at m/z 125 [M − 28 − 44 − H] − , as well as the loss of CH2 = CH2 and CO2, thus confirming that the compound is ethyl gallate (11) [32].
The deprotonated molecule ion at m/z 495 (7) was identified as digalloyl quinic acid. The second order spectrum shows the lost group galloyl [M − 152 − H] − providing the ion at m/z 343. The fragmentation ion at m/z 343 yielded other fragments: m/z 191, related to the elimination of a galloyl group [M − H − 152] − ; and the corresponding fragmentation at m/z 169 eliminating the quinic acid with the formation of deprotonated gallic acid [25].
Ellagic acid and its derivatives were also identified in the C. ferrea extract. The Compounds 13, 18, 19 and 20 showed the deprotonated molecule ion at m/z 301, 447, 461 and 469, respectively. The substances displayed UV spectra with λmax of 254 and 365 nm and fragmentation patterns characteristic of ellagic acid derivatives, with losses of 44 Da (CO2) and 28 Da (CO) [19]. The ions were detected from the precursor ion m/z 301 of the ellagic acid (18) The mono-dimensional NMR spectra of this compound (19) showed chemical shifts similar to Nono et al. [45], which is why it was identified as 3-O-methylellagic acid 4'-O-β-D-arabinopyranoside. The analyses of the UV spectra showed that another class of substances identified in the ethanolic extract from the bark of C. ferrea is flavan-3-ol (16 and 6, Table 1). We observed the trimer (16) [20], which are formed by the retro-Diels-Alder fragmentation mechanism [29].

Validation Method
Although the separation of compounds could be achieved in the analysis by HPLC/ESI-IT-MS, there was an extensive chromatographic band between Rt = 25 min and Rt = 38 min, probably due to the high degree of polymerization of compounds in the sample, which resulted in poor chromatographic resolution. Therefore, in the HPLC-DAD analysis, a new chromatographic elution gradient was designed in order to minimize this effect, to improve the peak's resolution and, consequently, the area calculation for the peaks. With this in mind, one alternative found was to replace Solvent B, which is why methanol was replaced by acetonitrile ( Figure 3). The quantification curves were obtained by nine dilutions of the stock solution in a concentration range of 500 to 1.95 µg·mL −1 for gallic acid and eight dilutions of the stock solution in a concentration range of 333 to 2.6 µg·mL −1 . Chromatographic peak areas were calculated for each concentration, and interpolated values were determined as a function of concentration by using linear regression. Both curves showed good linearity. The following r 2 values were obtained: r 2 of gallic acid = 0.9999 (regression curve: y = 34178x − 105370); and r 2 of ellagic acid = 1.0 (regression curve: y = 169688x − 1257.8). The method presented good linearity over the concentration range evaluated ( Table 2). The limit of detection (LOD) for gallic acid was 0.78 μg·mL −1 , and the limit of quantification (LOQ) was 2.35 μg·mL −1 . For ellagic acid, the LOD was calculated to be 0.46 μg·mL −1 , and the LOQ was 2.60 μg·mL −1 . The overall intraday variations in retention times for the standards were 0.59% for gallic acid and 0.76% for ellagic acid. Interday variations were less than 4.76% for gallic acid and less than 3.66% for ellagic acid. These results indicate that the method had good repeatability, low variability and, thus, good precision. Selectivity was evaluated by comparing the retention time and UV spectra of each standard reference compound with that of the peaks in the C. ferrea extract (Figure 3).
In order to determine the most practical method, we used two reference standards available on the market. From initial viewing of the chromatogram, it became evident that the main constituents were phenolic acid derivatives. Some derivatives were expressed as gallic acid, while other derivatives were expressed as ellagic. This analysis could be performed because the electronic UV spectra of the standards are practically superimposable with those of the analyzed substances at the monitored wavelength. The concentration of the compound's chromatographic peak in the extract of the stem bark of C. ferrea was estimated, and the quantitative analysis results are reported in Table 3. Table 3. Estimation of the contents of phenolic acid derivatives in ethanolic extract from the stem bark of C. ferrea, expressed by the use of linear regression data of gallic acid (GA) and ellagic acid (EA).

Compound
Concentration ± SD (µg·mL −1 ) Standard Gallic acid 17 The results presented in Table 3 are in agreement with the quantification parameters evaluated for the method. For the six quantified substances, the contents are higher than the LOQ, but are within the ranges of its respective standard calibration curve. Therefore, the results show that the method is sensitive enough for analysis of the gallic and ellagic acid derivatives identified in the ethanolic extract from the stem bark of C. ferrea.

Mutagenic Activity
Carcinogenicity and mutagenicity are among the toxicological endpoints that pose the greatest concern for human health; thus, they are the object of intense research activity and recognized regulatory testing methods [46]. Several reliable and widely used mutagenicity tests were considered for use in this investigation; for example, the micronucleus test and the single cell gel electrophoresis (comet) assay [47,48]. However, the Ames test is used worldwide as an initial screen to determine the mutagenic potential of new chemicals and drugs. The Salmonella and Ames tests are in vitro models of chemical carcinogenicity and consist of a range of bacterial strains, which together are sensitive to a large array of DNA-damaging agents [46,49]. The four S. typhimurium strains used in this assay were TA97a, TA98, TA100 and TA102. Strains TA97 and TA98 are used to detect mutagens that cause frameshift mutations, while the TA100 and TA102 strains detect mutagens that cause DNA base-pair substitutions [49,50]. Table 4 shows the mutagenic activity detected in the Ames test, both in the presence and absence of microsomal activation. The responses are expressed as revertants/plate, and the mutagenic ratio is in parentheses. The ethanolic extract had no detectable mutagenicity. Table 4. Mutagenic activity expressed by means and standard deviations of the number of revertants/plates and mutagenicity index (MI) (value given in brackets) in TA98, TA100, TA102 and TA97a of Salmonella typhimurium treated with different concentrations of ethanol extract from the stem bark of C. ferrea, with (+S9) and without (−S9) metabolic activation. Given that medicinal plants have been contributing to health globally, it is imperative to evaluate their safety in terms of the mutagenic potential of their constituent compounds over both the short and long term. Short-term tests that detect genetic damage have provided information on the carcinogenic risk to humans for various chemicals [49].
In Brazil, a large number of herbal extracts are used in folk medicine to treat various types of disease. C. ferrea is a medicinal plant that is extensively used. In the present study, mutagenic activity assays with Salmonella demonstrated that the ethanolic extract from the stem bark of C. ferrea is not mutagenic. The absence of a mutagenic effect is important to the traditional system of medicine, supports a growing interest in the pharmacological evaluation of this plant and is a positive step forward in determining the safe use of this plant.

Chemicals and Reagents
Trifluoroacetic acid (TFA), acetonitrile and methanol were of HPLC grade and purchased from the Tedia Company (Fairfield, OH, USA). All other reagents used were of analytical grade. The following standards were used for quantitative analysis: gallic acid, ellagic acid and ethyl gallate, which were purchased from Sigma-Aldrich (São Paulo, Brazil). Water was purified with a Milli-Q system (Millipore, Billerica, MA, USA). All solutions prepared for HPLC were filtered through a 0.22-µm GHP (hydrophilic polypropylene) filter (Waters, Milford, MA, USA) before use.

Extraction
Dried and powdered stem bark (1610.0 g) of C. ferrea was macerated, at room temperature, with ethanol (4×, 48 h). The solution was evaporated under vacuum to give 260.09 g (16.2%) of crude ethanolic extract from the stem bark.
The isolated compounds were analyzed and identified by NMR, UV and MS experiments and then compared with data from the literature. The 1 H-NMR and 13 C-NMR 1D and 1H-NMR 2D-NMR 13 C g-HMBC experiments were performed on a Bruker ® 300 MHz (7.0 T) nuclear magnetic resonance spectrometer. For sample preparation for the NMR experiments, dimethyl sulfoxide (DMSO-d6, Cambridge Isotope Laboratories, Inc., Andover, MA, USA) was used.
For the compounds (2, 11, 18), we opted to develop a methodology based on analysis by HPLC-DAD, in an attempt to enable direct identification of secondary metabolites using co-injection of authentic standards (Sigma-purity of 98.5%). For the chromatographic condition, see Section 2.7.
For the other compounds identified in the ethanolic extract from the stem bark of C. ferrea, the characterization and identification of the peaks are shown in the Table 1, where UV-Vis absorptions and electrospray ionization mass spectrometry in negative ion mode of all the compounds detected are reported and compared with the literature data.

HPLC/ESI-IT-MS Analyses
The ethanol extracts of C. ferrea (stem bark) were analyzed by on-line HPLC/ESI-IT-MS, using a SURVEYOR MS micro system coupled in-line to an LCQ Fleet ion-trap mass spectrometer (Thermo Scientific). HPLC separation was conducted on a Phenomenex (Torrance, CA, USA) Synergi Hydro RP18 (250 × 4.6 mm i.d.; 4 μm), at a flow rate of 0.8 mL·min −1 . The mobile phases utilized were H2O (A) and MeOH (B). In both of these phases, acetic acid (0.1%) was added. Gradient elution was applied for 75 min. The initial condition was 5% (B), which was increased until reaching 100% (B) at 70 min and then held at 100% (B) for an additional 5 min.
The column effluent was split into two by means of an in-line T junction, which sent it to both ESI-MS and UV-DAD; 80% was sent to the UV-DAD detector and 20% was analyzed by ESI-MS in negative ion mode with a Fleet LCQ Plus ion-trap instrument from Thermo Scientific. The capillary voltage was set at −20 kV, the spray voltage at −5 kV and the tube lens offset at 100 V. The sheath gas (nitrogen) flow rate was set to 80 (arbitrary units), and the auxiliary gas flow rate was set at 5 (arbitrary units). Data were acquired in MS scanning modes. The capillary temperature was 275 °C. Xcalibur 2.1 software (Thermo Scientific, Waltham, MA, USA) was used for data analysis.

FIA-ESI-IT-MS n Analyses
The ethanolic extract was dissolved in methanol and injected by Flow Injection Analysis (FIA) into the ESI source (flow rate of 10 µL·min −1 ). Each ion, corresponding to each peak of the LC-MS chromatogram, was analyzed in the negative ESI-MS mode. Nitrogen was used both as a drying gas, at a flow rate of 60 (arbitrary units), and as a nebulizing gas. The ion spray voltage was 5 kV, and the tube lens offset was −55 V. The nebulizer temperature was set to 275 °C, and a potential of −4 V was used on the capillary. Negative ion mass spectra were recorded in the m/z range of 150-2000. The first event was a full-scan mass spectrum to acquire data on ions in the m/z range. The second event was an MS/MS experiment in which data-dependent scanning was performed on deprotonated molecules of the compounds. The collision energy for MS/MS was adjusted to 20%-30% and an activation time of 30 ms.

Quantification by HPLC-DAD of the Ethanolic Extract Obtained from the Stem Bark
The ethanolic extract obtained from the stem bark of C. ferrea, as well as the standards, were analyzed on a high performance liquid chromatograph (HPLC).

Preparation of Samples and Standards for Analysis by HPLC-DAD
The extract (20 mg) was weighed in a 2.0-mL flask, dissolved in 0.5 mL of DMSO (dimethylsulfoxide), brought to volume with methanol and then placed in an ultrasonic bath for 10 min.
The ellagic acid standard (1.0 mg) was dissolved in 3 mL of methanol:DMSO (50:50 v/v). This stock solution was used to prepare the remaining ellagic acid solutions included in the calibration curve. The samples of standard gallic acid (0.5 g) were dissolved, separately, in 1 mL of methanol. Methanol was used for subsequent dilutions.
The standard solutions and the crude extract were filtered through a 0.22-µm PTFE (Millex ® ) filter before analysis by HPLC-DAD.

Identification of Peaks
To characterize the chemical constituents (2, 11 and 18) or isolated compounds from the ethanolic extract of C. ferrea stem bark (3, 13 and 19), the retention times of each peak in the HPLC-DAD chromatograms were analyzed, along with UV spectra data and mass fragmentation patterns, and they were compared with literature data. Compounds purified by chromatography revealed a purity of 95%-99%.

Quantitative Determination of Constituents
The quantification of substances 2, 3, 11, 13, 18 and 19 was done using the external standard method. The quantification of individual constituents was performed using a regression curve, each point in triplicate. Measurements were performed at 254 nm, which is the maximum absorbance.

Linearity, Detection Limit, Quantification Limit and Precision
The calibration curves were obtained based on eight (ellagic acid) and nine (gallic acid) levels of concentration of standard mixtures. Chromatogram peak areas at 254 nm were plotted against the known concentrations of the standard solutions, in order to establish the calibration equations. The stock solutions were dissolved separately in spectroscopy grade methanol in order to obtain solutions that were appropriately diluted for each of the substances to furnish the following concentrations: A linear least squares regression of the peak areas as a function of the concentrations was performed to determine the correlation coefficients. The equation parameters (slope and intercept) of each standard curve were used to obtain the concentration values for the samples. The LOD and LOQ were calculated based on the standard deviation of the y-intercept (σ) and the slope of the calibration curve (S), which were obtained from linear regression. The LOD was calculated using the expression 3.3 σ/S, and the LOQ was calculated using 10 σ/S. The precision was expressed as a relative standard deviation (RSD). The areas under curves and retention times of the three consecutive injections, performed at each concentration and on three different days, were used to calculate interday precision (% RSD). Intraday precision data for peak areas and retention times were calculated from six non-consecutive injections, performed at each concentration on the same day.

Salmonella/Microsome Assay
Mutagenicity assays were performed by pre-incubating test extracts for 20-30 min with the Salmonella typhimurium strains TA100, TA98, TA97a and TA102, either with or without metabolic activation [50]. S. typhimurium strains were kindly provided by Dr. B. Ames, University of California, Berkeley, CA, USA. The S9-mix was freshly prepared before each test, with an Aroclor-1254-induced rat liver fraction purchased (lyophilized) from Moltox (Molecular Toxicology Inc.). The metabolic activation system consisted of 4% of the S9 fraction, 1% of 0.4 M MgCl2, 1% of 1.65 M KCl, 0.5% of 1 M D-glucose-6-phosphate disodium, 4% of 0.1 M NADP, 50% of 0.2 M phosphate buffer and 39.5% sterile distilled water. Five different doses of the test extract (0.26, 0.52, 1.04, 1.56 and 2.08 mg/plate) were diluted with DMSO, the concentrations of which were selected based on a preliminary toxicity test. In all subsequent assays, the upper limit of the dose range was either the highest non-toxic dose or the lowest toxic dose determined in the preliminary assay. The toxicity was apparent either as a reduction in the number of His + revertants or as an alteration in the auxotrophic background (i.e., the background lawn). The various concentrations of tested compounds were added to 500 µL of buffer (pH 7.4) and 100 µL of bacterial culture and incubated at 37 °C for 20-30 min. Then, 2 mL of top agar were added to the mixture, which was subsequently poured onto a plate containing minimum agar. The plates were incubated at 37 °C for 48 h, and the His+ revertant colonies were counted manually. The influence of metabolic activation was tested by adding 500 µL of the S9 mixture in place of the buffer. All experiments were performed in triplicate.
The standard mutagens used as positive controls in experiments without the S9 mix were 4-nitro-o-phenylenediamine (10 µg/plate) for TA98 and TA97a, sodium azide (2.5 µg/plate) for TA100 and mitomycin C (3 µg/plate) for TA102. 2-Anthramine (0.125 µg/plate) was used in experiments with metabolic activation of all strains. DMSO served as the negative (solvent) control. The results were analyzed with the Salanal statistical software package (U.S. Environmental Protection Agency, Monitoring Systems Laboratory, Las Vegas, NV, version 1.0, from Research Triangle Institute, RTP, NC, USA) [51], adopting the model of Bernstein et al. [52]. The data (revertants/plate) were assessed by variance analysis (ANOVA), followed by linear regression. The mutagenic index (MI) was also calculated for each dose as the average number of revertants per plate divided by the average number of revertants per plate of the negative (solvent) control. A sample was considered to be positive when MI ≥ 2 for at least one of the tested doses and if the response was dose dependent [16,53].

Conclusions
In this study, a comprehensive identification of chemical composition was conducted using the HPLC/ESI-IT-MS and FIA-ESI-IT-MS n methods, with a total of twenty compounds being identified. Additionally, a quantitative validation method was developed for the simultaneous determination of the six compounds, which were derivatives of the ellagic acid and gallic acid. The absence of a mutagenic effect in the ethanolic extract from the stem bark of C. ferrea is important for traditional medicine, supports a growing interest in the pharmacological evaluation of this plant and is a positive step forward in determining the safe use of this species.