Revising Reverse-Phase Chromatographic Behavior for Efficient Differentiation of Both Positional and Geometrical Isomers of Dicaffeoylquinic Acids

Dicaffeoylquinic acids (diCQAs) are plant metabolites and undergo trans-cis-isomerization when exposed to UV irradiation. As such, diCQAs exist in both trans- and cis-configurations and amplify the already complex plant metabolome. However, analytical differentiation of these geometrical isomers using mass spectrometry (MS) approaches has proven to be extremely challenging. Exploring the chromatographic space to develop possible conditions that would aid in differentially separating and determining the elution order of these isomers is therefore imperative. In this study, simple chromatographic parameters, such as column chemistry (phenyl versus alkyl), mobile phase composition (methanol or acetonitrile), and column temperature, were investigated to aid in the separation of diCQA geometrical isomers. The high-performance liquid chromatography photodiode array (HPLC-PDA) chromatograms revealed four isomers post UV irradiation of diCQA authentic standards. The elution profile/order was seen to vary on different reverse-phase column chemistries (phenyl versus alkyl) using different mobile phase composition. Here, the elution profile/order on the phenyl-derived column matrices (with methanol as the mobile phase composition) was observed to be relatively reproducible as compared to the alkyl (C18) columns. Chromatographic resolution of diCQA geometrical isomers can be enhanced with an increase in column temperature. Lastly, the study highlights that chromatographic elution order/profile cannot be relied upon to fathom the complexity of isomeric plant metabolites.


Introduction
Dica eoylquinic acids (diCQAs) are plant secondary metabolites that are part of the family of bioactive metabolites called chlorogenic acids. Dica eoylquinic acids (diCQAs) are formed from an esteri cation reaction between quinic acid and 2 units of the hydroxycinnamic acid (HCA) derivative, ca eic acid [1,2]. It is reported that HCA derivatives such as ca eic acid are initially synthesized in the transcon guration through the phenylpropanoid pathway [3]. However, due to the 1,2-disubstituted alkenic molecular structure (carbon-carbon double bond), these molecules absorb light (UV light) at a speci c wavelength and readily convert to the cis-geometry [3][4][5][6][7].
approximately 120-fold more than the trans-form [8]. In addition, in vacuo studies have demonstrated that HCA derivatives, such as dica eoyltartaric acid (chicoric acid) and diCQA, possess anti-human immunode ciency virus (HIV) type 1 DNA integrase activity, with the biological activity attributed to their cis-isomers [10][11][12]. However, currently, only a few in vitro and in vivo studies, that show the biological activities of cis-isomers of HCA derivatives, exist.
is is possibly due to the lack of knowledge about the existence of cis-isomers of HCA derivatives or the lack of cisform commercial standards [3].
Authentic standards of most of the HCA derivatives (trans-isomers), such as diCQAs, are commercially available, and these standards can be used to produce their ciscounterparts through the process of photoisomerization [4,6,12,14,15]. Identi cation of these related geometrical compounds using analytical techniques such as liquid chromatography linked to mass spectrometry (LC-MS) has proven impossible as they produce similar/identical MS fragmentation patterns [4]. As such, these analytical challenges have driven e orts in exploring the chromatographic space to suggest possible conditions that would aid in differentially separating and identifying these isomers [4,12]. In this endeavor, Cli ord et al. UV irradiated ve di erent authentic standards of diCQA positional isomers, namely, 1,3-diCQA, 1,5-diCQA, 3,4-diCQA, 3,5-diCQA, and 4,5-diCQA, and analyzed the samples on a phenyl-hexyl column using aqueous acetonitrile as part of the mobile phase composition [4]. From the study, Cli ord et al. anticipated three possible cis-isomers for all ve of the diCQA positional isomers. e number of cis-isomers can be attributed to the stereochemistry of the quinic acid unit (positions 1, 3, 4, and 5) to which the ca eic acid units are esteri ed [4,14] (Scheme 1). As such, two asymmetrical mono-cis-isomers (resulting from cis-isomerization on the respective ca eoyl arms on the quinic acid unit) and one di-cis-isomer (resulting from the cis-geometry on both ca eoyl arms on the quinic acid unit) were anticipated for each diCQA positional isomer. However, in the study by Cli ord et al., only two (instead of three) cis-isomers were observed for 1,3-diCQA, 3,4-diCQA, and 3,5-diCQA positional isomers [4].
Fundamentally, column chemistry [12,16,17], mobile phase composition [12,16], and column temperature [18,19] are the essential factors in de ning chromatographic separation during LC analyses. For instance, under reversephase conditions, a phenyl-derived column matrix may produce a di erent elution pro le (i.e., number of isomers separated) compared to an alkyl-derived column matrix, due to the interactions involved in the retention mechanism of the respective columns [12,16].
e aims of the current study were to reproduce and expand on the results observed by Cli ord et al. [4]. In our study, the abovementioned positional isomers of diCQAs ( and 4,5-diCQA) were UV irradiated, and the resulting samples were analyzed on nine di erent column chemistries, ve phenyl-derived columns and four alkyl (C 18 ) columns. e e ect of chromatographic parameters (column choice, mobile phase composition, and column temperature) on the chromatographic separation of the UV-generated diCQA geometrical isomers was evaluated. Insight on the chromatographic elution order of these metabolites (both transand cis-isomers) will contribute to ongoing e orts in designing analytical methods for di erential identi cation of isomers contributing to plant sample dimensionality. Such chromatographic separation e orts will further allow collection (LC fractionation) of these peaks (metabolites) to study their respective bioactivity di erences.

UV Irradiation.
A 1 mg/mL solution of each trans-diCQA positional isomer was prepared with 100% methanol. Irradiation of samples was conducted following the procedure described elsewhere [15]. Brie y, the solution (for each positional isomer) was placed in a Spectroline UV lamp operating at 254 nm with an intensity of 390 μW/cm 2 . e lamp was not covered with any notch lter. Irradiation was conducted for four hours (4 h), and aliquots (100 µL) were taken at 0 h (before irradiation) and at 4 h post irradiation. e aliquots were diluted 10 × with 100% methanol. All the samples were placed in amber vials and subjected to HPLC-PDA analyses.

HPLC-PDA Parameters.
e HPLC system used was a Shimadzu SCL-10A VP (Kyoto, Japan), equipped with a PDA controlled by Shimadzu VP software v. 5.31. Column oven temperature was set at 30°C and 50°C. e injection volume was 3 µL. A binary solvent mixture was used, consisting of Milli-Q water (eluent A) containing 0.1% formic acid and methanol or acetonitrile (eluent B) containing 0.1% formic acid. e initial conditions were 10% B at a ow rate of 0.2 mL/min and were maintained for 1 min, followed by an increase to 40% B at 15 min; the conditions were maintained for 2 min, followed by multiple gradients to 90% at 20 min; and the conditions were kept constant for 3 min and then changed to the initial conditions (10% B) after 5 min, followed by a 7-min isocratic wash at 10% B to re-equilibrate the column. e total chromatographic run time was 35 min. e PDA detector scanning range was set from 220 to 400 nm, and the chromatograms were processed at 325 nm. A column temperature study was conducted using a longer LC program (45 mins) to enhance the separation of diCQA geometrical isomers. e column oven temperature was set at 30, 35, 40, 45, 50, 55, and 60°C.

Results and Discussion
In this study, positional isomers of diCQAs (Scheme 1) were UV irradiated, and the resulting samples were analyzed under reverse-phase chromatographic conditions using nine di erent column chemistries. Here, for each column, either methanol (containing 0.1% formic acid) or acetonitrile (containing 0.1% formic acid) was used as the mobile phase/eluent B, with the initial column oven temperature set at 30 and then 50°C. Following analyses of the results, the use of aqueous methanol as the mobile phase/eluent B showed enhanced separation and, as such, the results discussed herein will be those obtained with methanol as part of the mobile phase composition [12]. e enhanced chromatographic separation using methanol as part of the binary solvent mixture is due to the weak eluent nature of methanol, which enhances the chromatographic separation of aromatic compounds by promoting longer retention within the column [12,20]. However, the results obtained with aqueous acetonitrile will also be referred to when necessary. In addition, unless stated otherwise, the results discussed herein are those obtained with the column oven temperature set at 30°C. e chromatograms of the nonirradiated (0 h) and irradiated (4 h) samples were compared and the retention times (t R ) of the peak observed from the nonirradiated samples were used to identify the trans-isomers from each irradiated diCQA sample, for each column used to conduct the study (Figure 1). A summary of the chromatographic results is represented in Table 1, where the void volume/"dead time" was assessed for each column by the inspection of the chromatograms, and the resulting capacity factors (k) for the trans-and cis-isomers are shown.

e E ect of Column Chemistry on the Separation of diCQA Geometrical Isomers.
e results show that the chromatographic pro le for the UV-irradiated sample of 1,3-diCQA is consistent on both the bi-phenyl and C 18 column matrices (Figure 2(a)) and consistent with the results achieved by Cli ord et al. [4]. On both the bi-phenyl and C 18 column matrices (Figure 2(a)), two mono-cis-isomers (M * and M # ) were observed to elute after their respective trans-counterpart (T). According to Cli ord et al., a peak of minor intensity is considered the di-cis-isomer [4]; thus, from Figure 2(a), C was annotated as the di-cis-isomer. In a study by Zheng et al., where the 3,5-diCQA geometrical isomers were separated by ion mobility, the photoisomerization study revealed that the di-cis-isomer forms directly from both the mono-cis-isomers [15]. In this study, the di-cis-isomer was retained longer on both column matrices (bi-phenyl and C 18 ), suggesting the resolution of the di-cis-isomer from the other isomers ( Figure  2(a)). us, for simplicity, the elution order of the various isomers for 1,3-diCQA is referred to as TM * M # C, where for all diCQAs, T represents the di-trans, M * represents the rst eluting mono-cis-isomer, M # represents the second eluting mono-cis-isomer, and C represents the di-cis-isomer. Furthermore, the elution pro le/order for all diCQAs is summarized in Table 2.
For the UV-irradiated sample of 1,5-diCQA, four peaks (corresponding to the four isomers) were observed when using the bi-phenyl and C 18 column matrices, and it is apparent that the elution pro le/order di ers between the two column matrices (Figure 2(b)). On the C 18 column matrix, two peaks, the rst mono-cis-isomer (M * ) and the di-cis-isomer (C), elute before the trans-isomer (T) and the fourth peak, the second mono-cis-isomer (M # ), elutes after the di-trans-isomer, resulting in the elution order M * CTM # (Table 2). However, on the bi-phenyl column, the cis-isomers are seen to elute after the trans-isomer, resulting with the elution order TM * CM # (Figure 2(b)) ( Table 2). e elution order, TM * CM # , observed on the bi-phenyl columns was similar to the elution order observed by Cli ord et al., using a phenyl-hexyl column matrix [4]. e consistency observed on the phenyl-containing column matrices (bi-phenyl versus phenyl-hexyl) suggests the possible role of π-π interactions in the separation of these aromatic isomers [12,16,17,21]. In contrast, di erences in the elution order amongst the C 18 columns were observed ( Figure 3). Instead of the elution order M * CTM # seen in Figure 2(b) using a Raptor C 18 , the Ultra C 18 column produced the elution order CM * TM # (Figure 3). e elution order observed for the UV-irradiated standard of 3,4-diCQA demonstrated a relatively identical elution pro le, on columns showing enhanced separation of the four isomers. Where the rst monocis-isomer (M * ) elutes before the trans-isomer (T), the trans-isomer is followed by the second mono-cis-isomer (M # ) and lastly followed by the di-cis-isomer (C), resulting in the elution order M * TM # C (Figure 2(c)) ( Table 2). e similar elution order observed on both the bi-phenyl and C 18 column matrices can be attributed to the stereochemistry at positions 3 and 4 on the quinic acid unit (Scheme 1, Figure 2(c)). e similarities in the spatial arrangement of the ca eic acid units at these positions on the quinic acid unit could possibly result in comparable interactions with their surrounding environments (i.e., mobile and stationary phase). In addition, the similar elution pro le seen on both the bi-phenyl and C 18 column matrices for the UV-irradiated sample of 1,3-diCQA (TM * M # C) (Figure 2(a), Table 2) can also be attributed to the identical spatial arrangement at positions 1 and 3 on   for the UV-irradiated sample of 3,4-diCQA using a phenyl-hexyl column [4]. Here, the elution order observed was M * TM # instead of the elution order M * TM # C observed in our study and discussed above. Results observed by Cli ord et al. were also observed in our study when the phenyl-hexyl column was used with aqueous acetonitrile as the eluent (see Supplementary Figure S1), thus suggesting the possible coelution of the di-cis-isomer under these conditions. When the UV-generated geometrical isomers of 3,5-diCQA were analyzed on the phenyl-containing column matrices (at a column temperature of 30°C), only three peaks (instead of four peaks) were observed and resulted with the elution order TM * M # (Figure 2(d), Table 2), as observed by Cli ord et al. [4]. On the C 18 column matrices, the elution order CM * M # T was observed (Figure 2(d), Table 2).  18 column matrix. e gure shows di erences and similarities in the elution pro les, dependent on either the bi-phenyl or C 18 column matrix. T represents the di-trans-isomer, M * represents the rst eluting mono-cis-isomer, M # represents the second eluting mono-cis-isomer, and C represents the di-cis-isomer.   Although not commonly considered a key parameter in reverse-phase chromatography, high column temperatures have been shown to enhance the separation due to a decrease in the viscosity of the mobile phase. In addition, retention factors are dependent on the distribution coe cients (k d ) of the analytes, which are temperature dependent. An increase in temperature enhances the separation [19,22,23]. us, to achieve the separation of 3,5-diCQA geometrical isomers on the phenyl-containing column matrices, a column temperature of 50°C was introduced and resulted in the elution order TCM * M # (Figure 4) instead of the elution order TM * M # seen in Figure 2(d).
e consistency in the elution pro le observed on the phenyl column matrices (TM * M # ) can be attributed to π-π interactions which enhance separation in phenyl-containing columns [12,16,17,20,21], and the elution pro le di erences observed between column matrices (phenyl versus alkyl matrices) can be attributed to the di erences in spatial arrangements of the ca eoyl units at positions 3 and 5 on the quinic acid unit. For instance, the di erent spatial arrangements at these positions a ect when cis-isomers elute on a phenyl column matrix (elutes after the trans-isomer; TCM * M # ) versus an alkyl column (elutes before the trans-isomer; CM * M # T).
Finally, the elution order for the geometrical isomers of 4,5-diCQA was seen to be M * TCM # , only on three bi-phenyl columns and two C 18 columns as summarized in Table 2 and demonstrated in Figure 2(e). Interestingly, the other two C 18 columns showed a di erent elution order ( Figure 5); thus, instead of M * TCM # (Figure 5(a)), M * CTM # was observed ( Figure 5(b)). Furthermore, for this sample, Cli ord et al. observed the elution order TM * CM # when using a phenylhexyl column [4]. In our study, these results were also observed using a Phenomenex bi-phenyl column with acetonitrile as part of the mobile phase composition (Supplementary data, Figure S2). us, care must be taken when analyzing geometrical isomers of 4,5-diCQA on di erent column matrices.
Despite the inconsistent chromatographic elution pro les observed for the UV-irradiated sample of 4,5-diCQA, what seems to be consistent is the later elution of the second eluting mono-cis-isomer (M # ) ( Figure 5).
is is also evident in samples of 1,5-diCQA (Figure 2(b)) and 3,5-diCQA (Figure 2(d), Figure 4), especially when analyzed on phenyl-containing column matrices. is mono-cis-isomer could possibly be a cis-isomer at position 5 on the quinic acid unit. According to Cli ord et al., a cisgeometry at position 5 on the quinic acid for mono-acyl chlorogenic acids results in intramolecular hydrogen bonding at two positions; (1) the ca eoyl carbonyl (C�O) group and the 4-OH group on the quinic acid unit and (2) the ca eoyl 3′-OH group and the carbonyl group at position 1 on the quinic acid unit [4], thus rendering the molecules less hydrophilic due to unavailable hydroxyl groups and the compact nature of the molecule. In this study, it is uncertain to what extent the above applies to the diCQAs (Supplementary data, Figure S3).  Journal of Analytical Methods in Chemistry apparent that determining the elution order of diCQA geometrical isomers on di erent reverse-phase column matrices shows inconsistencies (Table 2). Within the C 18 column matrices, di erent chromatographic elution proles were observed, and the Ultra C 18 column showed the worst performance when analysis was conducted with the column temperature set at 30°C. As such, column temperature was varied (30-60°C) to enhance the separation of the diCQA geometrical isomers on the Ultra C 18 column ( Figure 6). From Figure 6, an increase in column temperature showed a positive e ect on the resolution of the geometrical isomers of 1,3-diCQA, 1,5-diCQA, 3,5-diCQA, and 4,5-diCQA. An increase in temperature resulted in the earlier elution of analytes and resolution of the UV-irradiated diCQA geometrical isomers ( Figure 6). Similar results were observed by Nguyen et al., whereby a pharmaceutical cocktail was chromatographically separated at temperatures 30°C and 90°C, and the temperature at 90°C enhanced the separation and decreased the analysis time [19]. Furthermore, in our study the UV-irradiated sample of 3,4-diCQA showed the separation and resolution of only three isomers, suggesting coelution of the di-cis-isomer. For this sample (UV-irradiated sample of 3,4-diCQA), the elution order (M * TM # ) was also observed using the phenyl-hexyl column coupled with aqueous acetonitrile as part of the mobile phase (Supplementary data, Figure S1). e results obtained in this temperature study suggest that some column matrices are incapable of separating/distinguishing all the available isomers in the sample even post optimization of the chromatographic parameters.

Conclusion
is study demonstrates that positional isomers of diCQA samples produce three cis-isomers post UV irradiation, and separation of these isomers is dependent on optimization of primary chromatographic parameters. As such, column chemistry, mobile phase composition, and column temperature in uence the chromatographic elution pro le of the structurally related compounds, thus hindering identi cation. From the above results, it is apparent that determining the elution pro le/order of diCQA geometrical isomers on di erent reverse-phase column matrices (phenyl versus alkyl) shows inconsistencies. However, a relatively consistent elution order was observed using the phenyl-containing column matrices, suggesting the important role of π-π interactions in the separation of the respective diCQA geometrical isomers. e results show di erent elution pro les between C 18 column matrices from di erent column suppliers, suggesting that column manufacturing is not standardized.
e study also shows that column temperature can be used to enhance the separation of the isomers. Furthermore, the number of observed isomers depends on the capability of the column to distinguish the isomers. For instance, using the Ultra C 18 column, the separation of the 3,4-diCQA geometrical isomers was enhanced by the introduction of column temperature; however, the di- cis-isomer was not observed on the chromatogram, suggesting coelution of the di-cis-isomer. Lastly, the study highlights that chromatographic elution order/pro le cannot be relied upon to fathom the complexity of isomeric plant metabolites and that more advanced analytical methods need to be developed to achieve this goal. Advancement in analytical approaches can include hyphenation of high-temperature liquid chromatography e gure shows that an increase in column temperature enhances the separation of diCQA geometrical isomers.