- Email: [email protected]
Comprehensive thin-layer chromatography mass spectrometry of flavanols from Juniperus communis L. and Punica granatum L.
Journal of Chromatography A, 1289 (2013) 119–126
Contents lists available at SciVerse ScienceDirect
Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma
Comprehensive thin-layer chromatography mass spectrometry of flavanols from Juniperus communis L. and Punica granatum L. Samo Smrke a , Irena Vovk a,b,∗ a b
National Institute of Chemistry, Laboratory for Food Chemistry, Hajdrihova 19, SI-1000 Ljubljana, Slovenia EN-FIST Centre of Excellence, Dunajska 156, SI-1000 Ljubljana, Slovenia
a r t i c l e
i n f o
Article history: Received 12 November 2012 Received in revised form 25 February 2013 Accepted 10 March 2013 Available online 16 March 2013 Keywords: HPTLC TLC–MS Flavanols Proanthocyanidins Juniper Pomegranate
a b s t r a c t The coupling of thin-layer chromatography with mass spectrometry (TLC–MS) for the analysis of monomeric flavanols and proanthocyanidins in samples presented as complex matrices has been studied. The elution conditions for TLC–MS were optimised and full scans were compared with selected reaction monitoring for the MS detection of compounds. The performance of silica gel and cellulose plates with different developing solvents in TLC–MS was assessed. Cellulose plates provided superior sensitivity while ionisation suppression was encountered with silica plates. The use of a HILIC guard column beyond the elution head was found to facilitate detection of monomer compounds on silica plates. A new comprehensive TLC × MS procedure for screening flavanols in the entire chromatogram was developed as an alternative to the use of 4-dimethylaminocinnamaldehyde to determine the locations of compounds on the plate. This new procedure was applied to detect flavanols in the peel of Punica granatum L. fruits and in seeds of Juniperus communis L., in which flavanols and proanthocyanidin dimers and trimers were detected for the first time. © 2013 Elsevier B.V. All rights reserved.
1. Introduction Recent trends in development of thin-layer chromatography (TLC) have been focused on its miniaturisation and hyphenation with mass spectrometry (MS) [1–5]. Although the resolution of TLC is greatly inferior to that of high-performance liquid chromatography (HPLC), some of its properties such as simplicity, economy, easy operation and low consumption of solvents have regenerated interest in TLC. Interfacing MS to TLC however is not as straightforward as with HPLC or gas chromatography. After TLC separation, analytes remain absorbed on the solid phase but there are various direct sampling methods and techniques available with which to couple TLC and MS [2,3]. Currently, there is focus in coupling TLC with ambient pressure ionisation MS [1] and one of the most common and possibly the simplest way to interface TLC–MS is by means of an elution head based interface. In 2009 a Luftman type elution head [6] has become commercially available and there are reports of its use in the analysis of food supplements and pharmaceuticals and in fingerprinting of herbal extracts [5,7–13]. Interestingly this TLC–MS interface has also found use as a sampling device for analysis of dried blood spots [14,15].
∗ Corresponding author at: National Institute of Chemistry, Hajdrihova 19, SI-1000 Ljubljana, Slovenia. Tel.: +386 1 4760 341; fax: +386 1 4760 300. E-mail address: [email protected] (I. Vovk). 0021-9673/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.chroma.2013.03.012
Increasing use of such an interface demands an extensive investigation of its potential and accordingly we applied the interface to TLC–MS analysis of monomeric flavanols and proanthocyanidins as an example of interesting natural occurring compounds which are commonly analysed as components of a complex matrix. Proanthocyanidins are a group of naturally occurring flavanol monomeric units linked through interflavan bonds to form oligomeric and polymeric proanthocyanidins. The interest in these compounds is due to their antioxidant [16] and free radical scavenging potentials [17] which offer potential health benefits for prevention of diseases triggered by oxidative stress (cardiovascular diseases [18,19], type 2 diabetes [20] and cancer [21]) when consumed as food or food supplements [22]. The proanthocyanidins are members of a huge family of isomeric compounds with complex structures and their separation and identification is very difficult [23], demanding use of all available analytical resources. TLC has been implemented as a tool for rapid screening [24–29] and for quantification [30–34] of flavanols. We analysed flavanols from extracts of juniper seeds (Juniperus communis L.) and pomegranate fruit (Punica granatum L.). Extracts of both plants have very complex matrices, flavonoids in juniper berries and ellagitannins in pomegranate fruit [35], which create a challenge for sample preparation and in such cases, TLC has a considerable advantage over HPLC. There is substantial interest in plant material as source of antioxidants with health benefits because of its high phenolic content [36–38] but little interest in proanthocyanidins has been evident. Publications have described (+)-catechin, (+)-gallocatechin and three types of proanthocyanidin
120
S. Smrke, I. Vovk / J. Chromatogr. A 1289 (2013) 119–126
dimers in pomegranate fruit [38–40]. No proanthocyanidins have yet been detected in juniper berries or seeds, although four proanthocyanidins dimers were found in juniper bark [41], catechin and unidentified proanthocyanidins in juniper needles [42]. The flavonoid composition of juniper berries [43] has been investigated, but no flavanols were found. One of the difficulties associated with the use of the TLC–MS interface is the exact positioning of the elution head with respect to a chromatographic band on the plate. In case of intensely coloured compounds, compounds fluorescing visual light or those that have high absorbance at 254 nm visual positioning of spots is possible when a TLC darkroom and plates with added fluorescent agent are used. Alternatively, the exact positions of analytes can be determined using densitometry [5], but if a compound’s absorbance or quantity is insufficient to show a visible spot on the plate or one wants to take advantage of the benefits of the great sensitivity of MS, a different approach is necessary. One approach is to use selective derivatisation reagents which can lead to very low detection levels. An example is 4-dimethylaminocinnamaldehyde (DMACA), which has been used as a detection reagent for flavanols. The visual detection limit of 2 ng and 20 ng for (−)-epicatechin and procyanidin B2 and the densitometric detection limits of 0.2 ng and 2 ng for (−)-epicatechin and procyanidin B2, respectively [32–34] have been reported. In comparison, the densitometric limit of detection without derivatisation has been reported as 35 ng for (−)-epicatechin [31]. In this research we used a second approach where we sampled the chromatogram continuously by means of comprehensive TLC–MS thus circumventing elution head positioning. 2. Experimental 2.1. Materials Ethyl acetate, formic acid, acetic acid, n-propanol, hydrochloric acid, ethanol, 4-dimethylaminocinnamaldehyde (DMACA) and toluene were obtained from Merck (Darmstadt, Germany), acetone from Sigma–Aldrich (Steinheim, Germany), methanol and acetonitrile from J. T. Baker (Deventer, The Netherlands). Chemicals were of analytical grade, except methanol, acetonitrile and toluene which were of HPLC grade. Standard of (−)-epicatechin was obtained from Sigma Aldrich, (+)-catechin from Carl Roth (Karlsruhe, Germany), procyanidin B2 and (−)-epigallocatechin from Extrasynthese (Genay, France). Bidistilled water was used. 2.2. Preparation of standard solutions Standard stock solutions (0.1 mg/mL) were prepared in methanol. Working solutions (10 g/mL and 1 g/mL) were prepared by diluting stock solutions 10 and 100-fold with methanol. All standard solutions were stored at −18 ◦ C. 2.3. Preparation of sample test solutions Pomegranates (P. granatum L.) were obtained from a local market. Dried juniper berries (J. communis L.) were obtained from Kotany (Wolkersdorf, Austria). Pomegranate peel was lyophilised, stored at −18 ◦ C and ground in a mortar with a pestle prior to extraction. Juniper berries were split into peel and seeds, seeds were frozen in liquid nitrogen and then ground in a mortar with a pestle. Aliquots of 500 mg and 250 mg of pomegranate peel and juniper seeds, respectively, were weighed and extracted three times using 20 mL of 70% aqueous acetone to obtain crude extracts. Samples were vortexed for 2 min, kept in an ultrasonic bath for 5 min,
then vortexed for 1 min and centrifuged for 5 min at 4000 rpm. Supernatants were collected and acetone was evaporated under reduced pressure. The remaining aqueous solutions were extracted by liquid–liquid partitioning with ethyl acetate (3 × 20 mL). The ethyl acetate phases were collected and evaporated under reduced pressure, until ca. 1 mL of solvent remained. This was transferred into volumetric flasks which were then filled to 2 mL with ethyl acetate. Prior to analysis, sample test solutions were filtered through a 0.45 m membrane filter [Millex-HV hydrophilic polyvinylidene difluoride (PVDF); Millipore, Billerica, MA, USA].
2.4. TLC TLC was performed on 20 cm × 10 cm HPTLC cellulose plates and HPTLC silica gel 60 plates from Merck (Darmstadt, Germany), which were cut into segments of 10 cm × 10 cm. Standards and samples were applied with a Linomat IV (Camag, Muttenz, Switzerland) at an application speed of 8 s/L, 10 mm from the left edge as 6 mm bands, 4 mm apart, and 10 mm from the bottom of the plate. Application volumes for sample solution were 5 L and 20 L for pomegranate peel extract and 10 L for juniper seed extract. Applications of standards were 100, 150 or 200 ng for identification purposes and 2 ng and 10 ng when measuring limits of detection. The plates were developed up to 7 cm in an unsaturated normal developing chamber (Camag) for 10 cm × 10 cm plates. The developing solvents were n-propanol–water–acetic acid (4:2:1, v/v/v) [27] and pure water [26,27] for cellulose plates and acetone–toluene–acetic acid (6:3:1, v/v/v) for silica plates [44]. After development, the plates were dried for 2 min in a stream of warm air and the samples applied to the leftmost and the rightmost sides of the plates were visualised using optimised DMACA (4-dimethylaminocinnamaldehyde) detection reagent according to the previously described procedure [32]. The reagent was prepared in a 200 mL volumetric flask by dissolving 60 mg of DMACA in ethanol, adding 13 mL of conc. HCl and filling with ethanol. The plates were documented using Digistore 2 Documentation System using white light illumination (Camag).
2.5. TLC–MS coupling and MS conditions A TLC–MS interface (Camag) with a 4 mm × 2 mm elution head was used for elution of compounds from the HPTLC plates into the LCQ system (Thermo Finnigan, San Jose, CA, USA). Methanol was used as an eluent for most of experiments at a flow rate optimised to between 0.1 mL/min and 0.4 mL/min; in all experiments 1% formic acid in methanol at a flow rate of 10 L/min was added to the effluent prior to injection into the LCQ system. In case of full scan experiments for elution from silica gel 60 HPTLC plates by methanol–acetonitrile (1:1, v/v), a HILIC Luna (Phenomenex, Torrance, CA, USA) guard column was mounted between TLC–MS interface and ion source. The mass spectrometer was optimised by means of direct injection of standards into MS. Spectra were acquired using an electrospray ionisation (ESI) ion source in negative ion mode. The spray voltage was set to 4 kV, capillary temperature 200 ◦ C, capillary voltage −38.8 V and tube lens offset −5 V. Sheath gas and auxiliary gas flow rates were 95 a.u. and 14 a.u. (arbitrary units), respectively. Scan time for a m/z 200–600 full MS scan was 0.52 s. Selected reaction monitoring (SRM) was done by fragmentation with 35% collision energy. The most intense fragment peaks (m/z 245 for flavanol monomers and m/z 425 for flavanol dimers) were observed.
S. Smrke, I. Vovk / J. Chromatogr. A 1289 (2013) 119–126
121
2.6. Positioning of the TLC–MS interface To determine the positions of standard compounds on the plates, additional tracks of standard solutions were applied to the left and right sides of each plate. Only the tracks with these additional tracks were derivatised with DMACA. From the RF values of analytes visible on these two derivatised tracks, their position on other, non-derivatised tracks was estimated. For the comprehensive TLC × MS analysis each sample was applied to the plate at least twice in adjacent positions to minimise any possible deviations between chromatograms during the development of the plate. Elution from the sorbent was performed with TLC–MS in an alternating order e.g. from tracks 2 and 3 (Fig. 1) every 4 mm to acquire the entire chromatogram. 3. Results and discussion We investigated the potential of Camag TLC–MS interface for use in the analysis of monomeric flavanols and proanthocyanidins in plant samples with complex matrices. First the methods were developed, optimised and evaluated on standard solutions and then analysis of sample test solutions was performed. 3.1. Effect of eluent flow, sorbent material and developing solvent on TLC–MS of flavanol standards Eluent flow was optimised by observing the elution peaks of 200 ng standards of (−)-epicatechin and procyanidin B2 applied to undeveloped silica gel and cellulose plates. Methanol, in which the analytes are generally soluble, was chosen as an eluent because it is a strong eluent on both cellulose and silica gel stationary phases and it is compatible with ESI-MS. Elution from silica gel was faster than from a cellulose layer, and based on measurements of peak area, height and baseline width (Table 1), and considering also eluent consumption, 0.2 mL/min was chosen as an optimal flow, for both monomeric and dimeric compounds. No significant difference was found in elution speed between monomer and dimer. The dead volume was determined to be 35 L. Effect of sorbent and developing solvent to MS signal in full scan mode was investigated using methanol as the eluent. Signals from standards developed on plates with all three developing solvents were compared to those from undeveloped plates (Table 2).
Fig. 1. A juniper sample test solution applied (application volume 10 L) on cellulose HPTLC plate, developed using n-propanol–water–acetic acid (4:2:1, v/v/v) after comprehensive TLC–MS analysis of tracks 2 and 3. Track 1 was derivatised using DMACA detection reagent. Arrows indicate the sampling order used in the TLC × MS procedure.
The ratio in sensitivity between monomer and dimer was approximately the same under all conditions and it was assumed that the effect of the developing solvent is general, and not specific to a single analyte. No signals were detected from silica plates developed with acetone–toluene–acetic acid (6:3:1, v/v/v), even though the peak width was smaller and the peaks eluted earlier, indicating more effective elution of analytes. The background signal was not high enough to overwhelm ions produced by standards,
Table 1 Peak intensity and width of 200 ng of (−)-epicatechin standard extracted by means of TLC–MS with different eluent flow rates from undeveloped cellulose and silica gel HPTLC plates. Flow rate (mL/min)
0.1 0.2 0.3 0.4
Cellulose plates
Silica plates
Peak height (counts)
Peak area (counts)
Peak width (min)
Elution time (min)
Peak height (counts)
Peak area (counts)
Peak width (min)
Elution time (min)
76 172 251 284
2725 4910 3685 3994
1.2 1.1 0.8 0.6
1.7 1.4 0.8 0.8
60 171 214 371
1184 2199 1922 2810
0.6 0.5 0.3 0.3
1.1 0.7 0.5 0.4
Table 2 Intensity of signals of 150 ng of (−)-epicatechin and procyanidin B2 standards extracted by means of TLC–MS from different undeveloped and developed HPTLC plates. HPTLC plate/developing solvent
Silica gel/none Silica gel/acetone–toluene–acetic acid (6:3:1, v/v/v) Cellulose/none Cellulose/water Cellulose/n-propanol–water–acetic acid (4:2:1, v/v/v)
Peak area (counts) (−)-Epicatechin
Procyanidin B2
963 Not detected 5221 1111 2075
1621 Not detected 8924 3005 3891
122
S. Smrke, I. Vovk / J. Chromatogr. A 1289 (2013) 119–126
Table 3 Limits of detection (LOD) of (−)-epicatechin and procyanidin B2 standards, developed on cellulose plates with water, when detected by full scan and SRM mode. Analyte
(−)-Epicatechin Procyanidin B2
LOD (ng) Full scan
SRM
17 6
5 2
and it was assumed that stationary phase impurities or dissolving of the plate binder under acidic developing conditions suppress the ionisation of compounds of interest, therefore no signal is observed. 3.2. Optimisation of MS detection mode: Single Reaction Monitoring (SRM) One approach to detect standard signals from developed silica gel HPTLC plates was to use SRM mode in the MS analysis. In this SRM mode, monomer or dimer parent ions were collected and fragmented with 35% collision energy and the most intense fragment peaks (m/z 245 for monomers and m/z 425 for dimers) were then observed. With this arrangement it was possible to detect flavanols on a developed silica gel plate and confirm the detection by observing the fragment pattern. A similar gain in detection sensitivity is observed on undeveloped silica gel and cellulose, both developed and undeveloped. The integral of MS signal from developed silica gel plates for 100 ng of procyanidin B2 was 0.05 × 106 counts compared to 1.0 × 106 counts for undeveloped plates. The limits of detection when sampled from cellulose plates developed with water are presented in Table 3. 3.3. Detection of monomeric flavanols on silica plates by TLC–MS with online separation using a HILIC column An alternative way to detect signals from analytes in the presence of strong ionisation inhibitors is to separate them. Because the matrix compounds eluting from silica gel plates are most likely very polar as they are well washed out with methanol and acetonitrile (data not shown) we used a HILIC guard cartridge column placed
Fig. 2. TLC–MS spectrum noise of a (+)-catechin standard (m/z 289) (a), eluted from a developed silica gel HPTLC plate using methanol and acetonitrile (1:2, v/v) and the (+)-catechin standard (b) eluted from the HPTLC plate under the same conditions and followed by a simple online separation on a HILIC guard column.
Fig. 3. Chromatograms of sample test solutions and standards, developed using water on a cellulose plate (a) and n-propanol–water–acetic acid (4:2:1, v/v/v) on a cellulose plate (b). Plates were derivatised using DMACA detection reagent. The tracks are juniper seed sample test solution, application volume 10 L (1), (-)-epicatechin (2), (+)-catechin (3), procyanidin B2 (4), (-)-epigallocatechin (5) and pomegranate sample test solution, application volume 5 L (6).
S. Smrke, I. Vovk / J. Chromatogr. A 1289 (2013) 119–126
123
Fig. 4. TLC–MS spectrum from a juniper seed extract of (a) co-eluting proanthocyanidin dimers (m/z 577) and trimers (m/z 865) eluted from a cellulose plate, developed with water, sampled at RF 0.40 and (b) a spectrum of flavonoids co-eluting with proanthocyanidin dimers and trimers eluted from a cellulose plate, developed with n-propanol–water–acetic acid (4:2:1, v/v/v), sampled at RF 0.71.
after the elution head, before the eluent entering the ion source for a very simple online separation. Methanol proved to be a too strong eluent on HILIC to accomplish any gain in signal, so acetonitrile was added to the eluent. Methanol–acetonitrile 1:1, 1:2 and 1:3 (v/v) were tested and 1:2 gained the best results (Fig. 2b), showing no signals of background peaks as observed when eluting without online separation (Fig. 2a). Unfortunately this approach was shown to work only on monomers, because they interact very little with HILIC stationary phase, as there was no MS signal observed when analysing procyanidin B2. This is probably due to procyanidin B2 having a far higher retention on the HILIC stationary phase, so it was not possible to separate it from interfering compounds. 3.4. TLC–MS of sample test solutions As described earlier, it was only possible to detect monomers on silica plates by means of TLC–MS and the use of these plates was therefore limited to this application. Cellulose HPTLC plates were far more interesting for use with TLC–MS as two different developing solvents, pure water and n-propanol–water–acetic acid (4:2:1, v/v/v), offer different separation performance. Methanol at a flow rate of 0.2 mL/min was used as the eluent and no additional HILIC separation was needed. If screening for only specific flavanols with known fragmentation pattern was needed, the preferred detection
Fig. 5. TLC–MS spectrum of pomegranate fruit peel sample test solution, showing proanthocyanidin dimers composed of two (+)-catechin units (m/z 577), one (+)catechin and one (+)-gallocatechin units (m/z 593) and two (+)-gallocatechin units (m/z 609).
Fig. 6. TLC × MS chromatograms of a juniper seed extract test solution (a) developed on cellulose HPTLC plates using water and (b) on n-propanol–water–acetic acid (4:2:1, v/v/v). Application volumes were 10 L.
124
S. Smrke, I. Vovk / J. Chromatogr. A 1289 (2013) 119–126
method would be SRM. Not all analytes present in analysed samples are available as standards to optimise their fragmentation pattern; therefore to broaden the range of detectable compounds only full scan MS detection was used. Because of lack of resolution on HPTLC plates, compared to HPLC, high selectivity of DMACA is welcome in case of co-elution with non-flavanols, but any overlap of peaks cannot be detected by a single developing solvent. A combination of different developing solvents [27] is needed to separate the most common monomers and proanthocyanidins. Although MS cannot distinguish between flavanol epimers, it can give information about degree of polymerisation of proanthocyanidins and in some cases the type of monomer unit bonding or types of monomers present. Sample test solution was developed using cellulose sorbent, with both developing solvents and derivatised using the DMACA detection reagent. When crude extracts were applied to the plates, interfering compounds of similar chemistry and abundant in high quantities, e.g. ellagic acid and ellagitannins [37] in pomegranate fruit, overloaded the stationary phase and produced distorted chromatograms. Furthermore, when crude extracts were applied to the HPTLC plates, it was possible by means of TLC–MS to detect only the most abundant flavanols (+)-catechin from juniper seeds and (+)-catechin and (+)-gallocatechin from pomegranate peel. Consequently, a sample preparation step was necessary and used for all analyses presented here. Chromatograms of sample test solutions (application volume: juniper seed sample 10 L, pomegranate fruit peel sample 5 L) were compared with those of (+)-catechin, (−)-epicatechin, procyanidin B2 and (−)-epigallocatechin (application mass 100 ng) (Fig. 3). The cellulose plates provide information about the epimers present in samples. Both developing solvents that were used (Fig. 3a and b) separate (+)-catechin from (−)-epicatechin. When water was used as a developing solvent (Fig. 3a) the most abundant peak detected in both samples was (+)-catechin and compared to (+)-catechin, only very small amounts of (−)-epicatechin were detected. A peak that could not be identified solely by TLC caused interference with (−)-epicatechin when n-propanol–water–acetic acid (4:2:1, v/v/v) was used as the developing solvent. The most intense peak of the juniper seed sample seen in the chromatogram developed on cellulose plate using water (Fig. 3a) was probed with TLC–MS. Co-elution of proanthocyanidin dimers ([M−H]− , m/z 577) and trimers ([M−H]− , m/z 865) with the main peak of (+)-catechin ([M−H]− m/z 289) was observed. All three flavanols were detected for the first time in juniper seeds. We were unable by means of TLC–MS to detect a monomer peak (m/z 289) when sampling at the RF position of the (−)-epicatechin standard. Consequently, we confirmed that (−)-epicatechin could be present only in very small quantities and that the peaks at m/z 577 and m/z 865 represent oligomers of (+)-catechin monomeric units (Fig. 4a). Since no (−)-epicatechin was found on the cellulose plate developed with water, the interfering peak seen on the cellulose plates developed with n-propanol–water–acetic acid (4:2:1, v/v/v) was tested (Fig. 3b). No monomers could be detected, but proanthocyanidin dimers and trimers were detected together with other flavonoids (Fig. 4b), such as quercetin (m/z 301) [43]. Proanthocyanidins in pomegranate fruit are composed of (+)-catechin and (+)-gallocatechin monomeric units [35]. The additional hydroxyl group in (+)-gallocatechin enables the distinction of these monomeric units by MS. By sampling the peak seen on cellulose plates developed with n-propanol–water–acetic acid (4:2:1, v/v/v) at the position of (−)-epicatechin standard (Fig. 3b) (+)-gallocatechin was detected ([M−H]− , m/z 305); there was no peak at m/z 289. The presence of (−)-epigallocatechin was excluded because both epimers are separated in the chromatogram (Fig. 3b). Dimers composed only of (+)-catechin ([M−H]− , m/z 577),
(+)-gallocatechin ([M−H]− , m/z 609) or mixed dimers ([M−H]− , m/z 593) could also be identified from the MS (Fig. 5). We attempted to detect proanthocyanidin trimers but were unsuccessful, superior separation capacity or sample preparation would be necessary.
Fig. 7. TLC × MS chromatograms of pomegranate fruit peel test solutions developed on cellulose HPTLC plates using water (a) or n-propanol–water–acetic acid (4:2:1, v/v/v) as developing solvent (b, c). Application volumes: 5 L (a, b), 20 L (c).
S. Smrke, I. Vovk / J. Chromatogr. A 1289 (2013) 119–126
3.5. Comprehensive TLC × MS of sample test solutions To detect analytes using TLC–MS, when their position on the plate is unknown, the application termed “comprehensive TLC × MS” is used. It is not possible to achieve such detection by sampling a chromatogram of only one application on the plate. There are two reasons for this; first, after the elution is complete and the elution valve is closed, there is still a significant pressure from the remaining mobile phase in the elution head. When the head is lifted from the plate, a small amount of mobile phase usually spills on the plate, wetting the stationary phase near the sampling position. Because the mobile phase must be a very strong eluent the chromatogram around the sampling position becomes distorted. Second, lifting the elution head from the plate frequently damages the adjacent stationary phase. Attempts to sample TLC–MS from previous elutions nearby proved to be unsuccessful because the damage to the stationary phase caused eluent to leak from the elution head. Based on the information obtained by sampling peaks as seen in the chromatograms derivatised with DMACA, we performed comprehensive TLC × MS as follows. Juniper seed sample test solutions (10 L) were developed on cellulose plates with both developing solvents and TLC × MS was performed between RF 0.19 and 0.60 in RF steps of 0.032 (Fig. 6a) when water was used as the eluent and between RF 0.61 and 0.84 in RF steps of 0.029 (Fig. 6b) when n-propanol–water–acetic acid (4:2:1, v/v/v) was used as the developing solvent. Overlapping peaks of (+)-catechin, proanthocyanidin dimers and trimers could be seen when the plates were developed with water. When a TLC × MS chromatogram (Fig. 6a) is compared to an image of the HPTLC plate derivatised with DMACA detection reagent (Fig. 3a, track 1), a very large difference in sensitivity between (+)-catechin and proanthocyanidin oligomers is observed. The oligomers could only barely be detected on the derivatised plate if the image is computer enhanced, but with the TLC × MS the signal height is approximately the same. Peak broadening detected by comprehensive TLC × MS might have occurred as a result of undersampling and deviation between two chromatograms where TLC × MS is performed, but the effect is insufficient to produce peak overlap. Sample test solutions of pomegranate peel extract developed on cellulose plates were probed by TLC × MS at RF between 0.34 and 0.66 in RF steps of 0.029 (Fig. 7a) when developed with water (application volume 5 L). Two different application volumes were used when developed with n-propanol–water–acetic acid (4:2:1, v/v/v), 5 L (RF between 0.57 and 0.93 in steps of 0.033, Fig. 7b) and 20 L (RF between 0.57 and 0.93 in steps of 0.033, Fig. 7c). When water was used as a developing solvent, separation but not baseline separation, between compounds was achieved. Use of n-propanol–water–acetic acid (4:2:1, v/v/v) as a developing solvent gave a very good separation between (+)-catechin and (+)-gallocatechin by TLC × MS, but no other flavanol peaks could be detected by means of TLC–MS. When a larger quantity of sample was applied to the HPTLC cellulose plate both monomers were still separated and dimers could be easily detected.
Alternatively, online separation must be employed to separate any possible ionisation inhibitors from the analytes, therefore for flavanol analysis we favoured the use of cellulose sorbent where sensitivity was the highest. Comprehensive TLC × MS offers a better view of the analysed TLC chromatogram and in addition to TLC–MS can provide information about the retention behaviour of analytes on TLC plates. For analysis of flavanols by TLC–MS or TLC × MS, separation on cellulose sorbent, developed with npropanol–water–acetic acid (4:2:1, v/v/v) or pure water, eluted with methanol at a flow rate of 0.2 mL/min without additional HILIC separation after elution and full scan MS detection is the method of choice. Acknowledgement This study was carried out with financial support from the Slovenian Research Agency and EN-FIST Centre of Excellence (Dunajska 156, SI-1000 Ljubljana, Slovenia). References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16]
[17] [18] [19] [20] [21]
[22]
[23] [24] [25] [26]
4. Conclusions We have shown that TLC–MS is a very versatile tool for identification and detection of compounds on TLC plates. It is simple and provides valuable information about analytes even if there is a very specific and sensitive derivatisation reagent available, as is the case with flavanols. If there are problems with TLC–MS sensitivity when analytes are eluted from silica gel sorbents, the SRM mode can be applied to detect a given compound.
125
[27] [28] [29] [30] [31] [32] [33] [34] [35]
S.C. Cheng, M.Z. Huang, J. Shiea, J. Chromatogr. A 1218 (2011) 2700. G. Morlock, W. Schwack, J. Chromatogr. A 1217 (2010) 6600. G. Morlock, W. Schwack, Trends Anal. Chem. 29 (2010) 1157. S.P. Pasilis, G.J. Van Berkel, J. Chromatogr. A 1217 (2010) 3955. ´ B. Simonovska, A. Albreht, D. Agbaba, J. Chromatogr. A 1218 I. Vovk, G. Popovic, (2011) 3089. H. Luftmann, Anal. Bioanal. Chem. 378 (2004) 964. ´ D. Staszek, M. Waksmundzka-Hajnos, T. KowalM. Sajewicz, Ł. Wojtal, M. Natic, ska, J. Liq. Chromatogr. Relat. Technol. 34 (2011) 848. ´ Ł. Wojtal, M. Waksmundzka-Hajnos, T. KowalM. Sajewicz, D. Staszek, M. Natic, ska, J. Liq. Chromatogr. Relat. Technol. 34 (2011) 864. M. Sajewicz, Ł. Wojtal, M. Hajnos, M. Waksmundzka-Hajnos, T. Kowalska, J. Planar Chromatogr. 23 (2010) 270. M. Sajewicz, D. Staszek, M. Natic, M. Waksmundzka-Hajnos, T. Kowalska, J. Chromatogr. Sci. 49 (2011) 560. ´ B. Simonovska, A. Albreht, I. Vovk, J. Chromatogr. A 1231 (2012) Z. Rodic, 59. V. Glavnik, B. Simonovska, A. Albreht, I. Vovk, J. Planar Chromatogr. 25 (2012) 251. K. Naumoska, B. Simonovska, A. Albreht, I. Vovk, J. Planar Chromatogr. 26 (2013) 125. P. Abu-Rabie, N. Spooner, Anal. Chem. 81 (2009) 10275. K. Heinig, T. Wirz, A. Gajate-Perez, Bioanalysis 2 (2010) 1873. ˇ V. Koleˇckáˇr, Z. Reháková, E. Brojerová, K. Kuˇca, D. Jun, K. Macáková, L. Opletal, P. Draˇsar, L. Jahodáˇr, J. Chlebek, L. Cahlíková, Chem. Listy 106 (2012) 113. I. Spranger, B. Sun, A.M. Mateus, V. de Freitas, J.M. Ricardo-da-Silva, Food Chem. 108 (2008) 519. T. Kurosawa, F. Itoh, A. Nozaki, Y. Nakano, S. Katsuda, N. Osakabe, H. Tsubone, K. Kondo, H. Itakura, Atherosclerosis 179 (2005) 237. T.P. Kenny, C.L. Keen, P. Jones, H.J. Kung, H.H. Schmitz, M.E. Gershwin, Exp. Biol. Med. 229 (2004) 255. M. Tomaru, H. Takano, N. Osakabe, A. Yasuda, K. Inoue, R. Yanagisawa, T. Ohwatari, H. Uematsu, Nutrition 23 (2007) 351. D. Ramljak, L.J. Romanczyk, L.J. Metheny-Barlow, N. Thompson, V. Knezevic, M. Galperin, A. Ramesh, R.B. Dickson, Mol. Cancer Ther. 4 (2005) 537. H.A. Weber, A.E. Hodges, J.R. Guthrie, B.M. O‘Brien, D. Robaugh, A.P. Clark, R.K. Harris, J.W. Algaier, C.S. Smith, J. Agric. Food Chem. 55 (2007) 148. J. Valls, S. Millán, M.P. Martí, E. Borràs, L. Arola, J. Chromatogr. A 1216 (2009) 7143. I. Fecka, W. Cisowski, M. Luczkiewicz, J. Planar Chromatogr. 14 (2001) 405. I. Vovk, B. Simonovska, P. Vuorela, H. Vuorela, J. Planar Chromatogr. 15 (2002) 433. I. Vovk, B. Simonovska, S. Andrenˇsek, H. Vuorela, P. Vuorela, J. Chromatogr. A 991 (2003) 267. I. Vovk, B. Simonovska, H. Vuorela, J. Chromatogr. A 1077 (2005) 188. R. Nowak, J. Planar Chromatogr. 20 (2007) 43. E. Reich, A. Schibli, V. Widmer, R. Jorns, E. Wolfram, A. DeBatt, J. Liq. Chromatogr. Rel. Technol. 29 (2006) 2141. T. Hofmann, L. Albert, T. Rétfalvi, J. Planar Chromatogr. 17 (2004) 350. L. Pobłocka-Olech, M. Krauze-Baranowska, M. Wiwart, J. Planar Chromatogr. 20 (2007) 61. V. Glavnik, B. Simonovska, I. Vovk, J. Chromatogr. A 1216 (2009) 4485. V. Glavnik, B. Simonovska, I. Vovk, J. Planar Chromatogr. 23 (2010) 230. ´ D. Aˇsperger, S. Babic, ´ V. Glavnik, B. Simonovska, I. Vovk, D. Mutavdˇzic´ Pavlovic, J. Planar Chromatogr. 24 (2011) 482. C. Ben Nasr, N. Ayed, M. Metche, Z. Lebens. Unters. Forsch. 203 (1996) 374.
126
S. Smrke, I. Vovk / J. Chromatogr. A 1289 (2013) 119–126
[36] N. Miceli, A. Trovato, P. Dugo, F. Cacciola, P. Donato, A. Marino, V. Bellinghieri, T.M. La Barbera, A. Güvenc¸, M.F. Taviano, J. Agric. Food Chem. 57 (2009) 6570. [37] N.P. Seeram, L.S. Adams, S.M. Henning, Y. Niu, Y. Zhang, M.G. Nair, D. Heber, J. Nutr. Biochem. 16 (2005) 360. [38] G.W. Plumb, S. de Pascual-Teresa, C. Santos-Buelga, J.C. Rivas-Gonzalo, G. Williamson, Redox Rep. 7 (2002) 41. [39] S. de Pascual-Teresa, C. Santos-Buelga, J.C. Rivas-Gonzalo, J. Agric. Food Chem. 48 (2000) 5331.
[40] S. de Pascual-Teresa, J.C. Rivas-Gonzalo, C. Santos-Buelga, Int. J. Food Sci. Technol. 35 (2000) 33. [41] R. Engelshowe, Planta Med. 49 (1983) 170. [42] F. Martz, R. Peltola, S. Fontanay, R.E. Duval, R. Julkunen-Tiitto, S. Stark, J. Agric. Food Chem. 57 (2009) 9575. [43] M. Innocenti, M. Michelozzi, C. Giaccherini, F. Ieri, F.F. Vincieri, N. Mulinacci, J. Agric. Food Chem. 55 (2007) 6596. [44] A.G.H. Lea, J. Sci. Food Agric. 29 (1978) 471.
Contents lists available at SciVerse ScienceDirect
Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma
Comprehensive thin-layer chromatography mass spectrometry of flavanols from Juniperus communis L. and Punica granatum L. Samo Smrke a , Irena Vovk a,b,∗ a b
National Institute of Chemistry, Laboratory for Food Chemistry, Hajdrihova 19, SI-1000 Ljubljana, Slovenia EN-FIST Centre of Excellence, Dunajska 156, SI-1000 Ljubljana, Slovenia
a r t i c l e
i n f o
Article history: Received 12 November 2012 Received in revised form 25 February 2013 Accepted 10 March 2013 Available online 16 March 2013 Keywords: HPTLC TLC–MS Flavanols Proanthocyanidins Juniper Pomegranate
a b s t r a c t The coupling of thin-layer chromatography with mass spectrometry (TLC–MS) for the analysis of monomeric flavanols and proanthocyanidins in samples presented as complex matrices has been studied. The elution conditions for TLC–MS were optimised and full scans were compared with selected reaction monitoring for the MS detection of compounds. The performance of silica gel and cellulose plates with different developing solvents in TLC–MS was assessed. Cellulose plates provided superior sensitivity while ionisation suppression was encountered with silica plates. The use of a HILIC guard column beyond the elution head was found to facilitate detection of monomer compounds on silica plates. A new comprehensive TLC × MS procedure for screening flavanols in the entire chromatogram was developed as an alternative to the use of 4-dimethylaminocinnamaldehyde to determine the locations of compounds on the plate. This new procedure was applied to detect flavanols in the peel of Punica granatum L. fruits and in seeds of Juniperus communis L., in which flavanols and proanthocyanidin dimers and trimers were detected for the first time. © 2013 Elsevier B.V. All rights reserved.
1. Introduction Recent trends in development of thin-layer chromatography (TLC) have been focused on its miniaturisation and hyphenation with mass spectrometry (MS) [1–5]. Although the resolution of TLC is greatly inferior to that of high-performance liquid chromatography (HPLC), some of its properties such as simplicity, economy, easy operation and low consumption of solvents have regenerated interest in TLC. Interfacing MS to TLC however is not as straightforward as with HPLC or gas chromatography. After TLC separation, analytes remain absorbed on the solid phase but there are various direct sampling methods and techniques available with which to couple TLC and MS [2,3]. Currently, there is focus in coupling TLC with ambient pressure ionisation MS [1] and one of the most common and possibly the simplest way to interface TLC–MS is by means of an elution head based interface. In 2009 a Luftman type elution head [6] has become commercially available and there are reports of its use in the analysis of food supplements and pharmaceuticals and in fingerprinting of herbal extracts [5,7–13]. Interestingly this TLC–MS interface has also found use as a sampling device for analysis of dried blood spots [14,15].
∗ Corresponding author at: National Institute of Chemistry, Hajdrihova 19, SI-1000 Ljubljana, Slovenia. Tel.: +386 1 4760 341; fax: +386 1 4760 300. E-mail address: [email protected] (I. Vovk). 0021-9673/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.chroma.2013.03.012
Increasing use of such an interface demands an extensive investigation of its potential and accordingly we applied the interface to TLC–MS analysis of monomeric flavanols and proanthocyanidins as an example of interesting natural occurring compounds which are commonly analysed as components of a complex matrix. Proanthocyanidins are a group of naturally occurring flavanol monomeric units linked through interflavan bonds to form oligomeric and polymeric proanthocyanidins. The interest in these compounds is due to their antioxidant [16] and free radical scavenging potentials [17] which offer potential health benefits for prevention of diseases triggered by oxidative stress (cardiovascular diseases [18,19], type 2 diabetes [20] and cancer [21]) when consumed as food or food supplements [22]. The proanthocyanidins are members of a huge family of isomeric compounds with complex structures and their separation and identification is very difficult [23], demanding use of all available analytical resources. TLC has been implemented as a tool for rapid screening [24–29] and for quantification [30–34] of flavanols. We analysed flavanols from extracts of juniper seeds (Juniperus communis L.) and pomegranate fruit (Punica granatum L.). Extracts of both plants have very complex matrices, flavonoids in juniper berries and ellagitannins in pomegranate fruit [35], which create a challenge for sample preparation and in such cases, TLC has a considerable advantage over HPLC. There is substantial interest in plant material as source of antioxidants with health benefits because of its high phenolic content [36–38] but little interest in proanthocyanidins has been evident. Publications have described (+)-catechin, (+)-gallocatechin and three types of proanthocyanidin
120
S. Smrke, I. Vovk / J. Chromatogr. A 1289 (2013) 119–126
dimers in pomegranate fruit [38–40]. No proanthocyanidins have yet been detected in juniper berries or seeds, although four proanthocyanidins dimers were found in juniper bark [41], catechin and unidentified proanthocyanidins in juniper needles [42]. The flavonoid composition of juniper berries [43] has been investigated, but no flavanols were found. One of the difficulties associated with the use of the TLC–MS interface is the exact positioning of the elution head with respect to a chromatographic band on the plate. In case of intensely coloured compounds, compounds fluorescing visual light or those that have high absorbance at 254 nm visual positioning of spots is possible when a TLC darkroom and plates with added fluorescent agent are used. Alternatively, the exact positions of analytes can be determined using densitometry [5], but if a compound’s absorbance or quantity is insufficient to show a visible spot on the plate or one wants to take advantage of the benefits of the great sensitivity of MS, a different approach is necessary. One approach is to use selective derivatisation reagents which can lead to very low detection levels. An example is 4-dimethylaminocinnamaldehyde (DMACA), which has been used as a detection reagent for flavanols. The visual detection limit of 2 ng and 20 ng for (−)-epicatechin and procyanidin B2 and the densitometric detection limits of 0.2 ng and 2 ng for (−)-epicatechin and procyanidin B2, respectively [32–34] have been reported. In comparison, the densitometric limit of detection without derivatisation has been reported as 35 ng for (−)-epicatechin [31]. In this research we used a second approach where we sampled the chromatogram continuously by means of comprehensive TLC–MS thus circumventing elution head positioning. 2. Experimental 2.1. Materials Ethyl acetate, formic acid, acetic acid, n-propanol, hydrochloric acid, ethanol, 4-dimethylaminocinnamaldehyde (DMACA) and toluene were obtained from Merck (Darmstadt, Germany), acetone from Sigma–Aldrich (Steinheim, Germany), methanol and acetonitrile from J. T. Baker (Deventer, The Netherlands). Chemicals were of analytical grade, except methanol, acetonitrile and toluene which were of HPLC grade. Standard of (−)-epicatechin was obtained from Sigma Aldrich, (+)-catechin from Carl Roth (Karlsruhe, Germany), procyanidin B2 and (−)-epigallocatechin from Extrasynthese (Genay, France). Bidistilled water was used. 2.2. Preparation of standard solutions Standard stock solutions (0.1 mg/mL) were prepared in methanol. Working solutions (10 g/mL and 1 g/mL) were prepared by diluting stock solutions 10 and 100-fold with methanol. All standard solutions were stored at −18 ◦ C. 2.3. Preparation of sample test solutions Pomegranates (P. granatum L.) were obtained from a local market. Dried juniper berries (J. communis L.) were obtained from Kotany (Wolkersdorf, Austria). Pomegranate peel was lyophilised, stored at −18 ◦ C and ground in a mortar with a pestle prior to extraction. Juniper berries were split into peel and seeds, seeds were frozen in liquid nitrogen and then ground in a mortar with a pestle. Aliquots of 500 mg and 250 mg of pomegranate peel and juniper seeds, respectively, were weighed and extracted three times using 20 mL of 70% aqueous acetone to obtain crude extracts. Samples were vortexed for 2 min, kept in an ultrasonic bath for 5 min,
then vortexed for 1 min and centrifuged for 5 min at 4000 rpm. Supernatants were collected and acetone was evaporated under reduced pressure. The remaining aqueous solutions were extracted by liquid–liquid partitioning with ethyl acetate (3 × 20 mL). The ethyl acetate phases were collected and evaporated under reduced pressure, until ca. 1 mL of solvent remained. This was transferred into volumetric flasks which were then filled to 2 mL with ethyl acetate. Prior to analysis, sample test solutions were filtered through a 0.45 m membrane filter [Millex-HV hydrophilic polyvinylidene difluoride (PVDF); Millipore, Billerica, MA, USA].
2.4. TLC TLC was performed on 20 cm × 10 cm HPTLC cellulose plates and HPTLC silica gel 60 plates from Merck (Darmstadt, Germany), which were cut into segments of 10 cm × 10 cm. Standards and samples were applied with a Linomat IV (Camag, Muttenz, Switzerland) at an application speed of 8 s/L, 10 mm from the left edge as 6 mm bands, 4 mm apart, and 10 mm from the bottom of the plate. Application volumes for sample solution were 5 L and 20 L for pomegranate peel extract and 10 L for juniper seed extract. Applications of standards were 100, 150 or 200 ng for identification purposes and 2 ng and 10 ng when measuring limits of detection. The plates were developed up to 7 cm in an unsaturated normal developing chamber (Camag) for 10 cm × 10 cm plates. The developing solvents were n-propanol–water–acetic acid (4:2:1, v/v/v) [27] and pure water [26,27] for cellulose plates and acetone–toluene–acetic acid (6:3:1, v/v/v) for silica plates [44]. After development, the plates were dried for 2 min in a stream of warm air and the samples applied to the leftmost and the rightmost sides of the plates were visualised using optimised DMACA (4-dimethylaminocinnamaldehyde) detection reagent according to the previously described procedure [32]. The reagent was prepared in a 200 mL volumetric flask by dissolving 60 mg of DMACA in ethanol, adding 13 mL of conc. HCl and filling with ethanol. The plates were documented using Digistore 2 Documentation System using white light illumination (Camag).
2.5. TLC–MS coupling and MS conditions A TLC–MS interface (Camag) with a 4 mm × 2 mm elution head was used for elution of compounds from the HPTLC plates into the LCQ system (Thermo Finnigan, San Jose, CA, USA). Methanol was used as an eluent for most of experiments at a flow rate optimised to between 0.1 mL/min and 0.4 mL/min; in all experiments 1% formic acid in methanol at a flow rate of 10 L/min was added to the effluent prior to injection into the LCQ system. In case of full scan experiments for elution from silica gel 60 HPTLC plates by methanol–acetonitrile (1:1, v/v), a HILIC Luna (Phenomenex, Torrance, CA, USA) guard column was mounted between TLC–MS interface and ion source. The mass spectrometer was optimised by means of direct injection of standards into MS. Spectra were acquired using an electrospray ionisation (ESI) ion source in negative ion mode. The spray voltage was set to 4 kV, capillary temperature 200 ◦ C, capillary voltage −38.8 V and tube lens offset −5 V. Sheath gas and auxiliary gas flow rates were 95 a.u. and 14 a.u. (arbitrary units), respectively. Scan time for a m/z 200–600 full MS scan was 0.52 s. Selected reaction monitoring (SRM) was done by fragmentation with 35% collision energy. The most intense fragment peaks (m/z 245 for flavanol monomers and m/z 425 for flavanol dimers) were observed.
S. Smrke, I. Vovk / J. Chromatogr. A 1289 (2013) 119–126
121
2.6. Positioning of the TLC–MS interface To determine the positions of standard compounds on the plates, additional tracks of standard solutions were applied to the left and right sides of each plate. Only the tracks with these additional tracks were derivatised with DMACA. From the RF values of analytes visible on these two derivatised tracks, their position on other, non-derivatised tracks was estimated. For the comprehensive TLC × MS analysis each sample was applied to the plate at least twice in adjacent positions to minimise any possible deviations between chromatograms during the development of the plate. Elution from the sorbent was performed with TLC–MS in an alternating order e.g. from tracks 2 and 3 (Fig. 1) every 4 mm to acquire the entire chromatogram. 3. Results and discussion We investigated the potential of Camag TLC–MS interface for use in the analysis of monomeric flavanols and proanthocyanidins in plant samples with complex matrices. First the methods were developed, optimised and evaluated on standard solutions and then analysis of sample test solutions was performed. 3.1. Effect of eluent flow, sorbent material and developing solvent on TLC–MS of flavanol standards Eluent flow was optimised by observing the elution peaks of 200 ng standards of (−)-epicatechin and procyanidin B2 applied to undeveloped silica gel and cellulose plates. Methanol, in which the analytes are generally soluble, was chosen as an eluent because it is a strong eluent on both cellulose and silica gel stationary phases and it is compatible with ESI-MS. Elution from silica gel was faster than from a cellulose layer, and based on measurements of peak area, height and baseline width (Table 1), and considering also eluent consumption, 0.2 mL/min was chosen as an optimal flow, for both monomeric and dimeric compounds. No significant difference was found in elution speed between monomer and dimer. The dead volume was determined to be 35 L. Effect of sorbent and developing solvent to MS signal in full scan mode was investigated using methanol as the eluent. Signals from standards developed on plates with all three developing solvents were compared to those from undeveloped plates (Table 2).
Fig. 1. A juniper sample test solution applied (application volume 10 L) on cellulose HPTLC plate, developed using n-propanol–water–acetic acid (4:2:1, v/v/v) after comprehensive TLC–MS analysis of tracks 2 and 3. Track 1 was derivatised using DMACA detection reagent. Arrows indicate the sampling order used in the TLC × MS procedure.
The ratio in sensitivity between monomer and dimer was approximately the same under all conditions and it was assumed that the effect of the developing solvent is general, and not specific to a single analyte. No signals were detected from silica plates developed with acetone–toluene–acetic acid (6:3:1, v/v/v), even though the peak width was smaller and the peaks eluted earlier, indicating more effective elution of analytes. The background signal was not high enough to overwhelm ions produced by standards,
Table 1 Peak intensity and width of 200 ng of (−)-epicatechin standard extracted by means of TLC–MS with different eluent flow rates from undeveloped cellulose and silica gel HPTLC plates. Flow rate (mL/min)
0.1 0.2 0.3 0.4
Cellulose plates
Silica plates
Peak height (counts)
Peak area (counts)
Peak width (min)
Elution time (min)
Peak height (counts)
Peak area (counts)
Peak width (min)
Elution time (min)
76 172 251 284
2725 4910 3685 3994
1.2 1.1 0.8 0.6
1.7 1.4 0.8 0.8
60 171 214 371
1184 2199 1922 2810
0.6 0.5 0.3 0.3
1.1 0.7 0.5 0.4
Table 2 Intensity of signals of 150 ng of (−)-epicatechin and procyanidin B2 standards extracted by means of TLC–MS from different undeveloped and developed HPTLC plates. HPTLC plate/developing solvent
Silica gel/none Silica gel/acetone–toluene–acetic acid (6:3:1, v/v/v) Cellulose/none Cellulose/water Cellulose/n-propanol–water–acetic acid (4:2:1, v/v/v)
Peak area (counts) (−)-Epicatechin
Procyanidin B2
963 Not detected 5221 1111 2075
1621 Not detected 8924 3005 3891
122
S. Smrke, I. Vovk / J. Chromatogr. A 1289 (2013) 119–126
Table 3 Limits of detection (LOD) of (−)-epicatechin and procyanidin B2 standards, developed on cellulose plates with water, when detected by full scan and SRM mode. Analyte
(−)-Epicatechin Procyanidin B2
LOD (ng) Full scan
SRM
17 6
5 2
and it was assumed that stationary phase impurities or dissolving of the plate binder under acidic developing conditions suppress the ionisation of compounds of interest, therefore no signal is observed. 3.2. Optimisation of MS detection mode: Single Reaction Monitoring (SRM) One approach to detect standard signals from developed silica gel HPTLC plates was to use SRM mode in the MS analysis. In this SRM mode, monomer or dimer parent ions were collected and fragmented with 35% collision energy and the most intense fragment peaks (m/z 245 for monomers and m/z 425 for dimers) were then observed. With this arrangement it was possible to detect flavanols on a developed silica gel plate and confirm the detection by observing the fragment pattern. A similar gain in detection sensitivity is observed on undeveloped silica gel and cellulose, both developed and undeveloped. The integral of MS signal from developed silica gel plates for 100 ng of procyanidin B2 was 0.05 × 106 counts compared to 1.0 × 106 counts for undeveloped plates. The limits of detection when sampled from cellulose plates developed with water are presented in Table 3. 3.3. Detection of monomeric flavanols on silica plates by TLC–MS with online separation using a HILIC column An alternative way to detect signals from analytes in the presence of strong ionisation inhibitors is to separate them. Because the matrix compounds eluting from silica gel plates are most likely very polar as they are well washed out with methanol and acetonitrile (data not shown) we used a HILIC guard cartridge column placed
Fig. 2. TLC–MS spectrum noise of a (+)-catechin standard (m/z 289) (a), eluted from a developed silica gel HPTLC plate using methanol and acetonitrile (1:2, v/v) and the (+)-catechin standard (b) eluted from the HPTLC plate under the same conditions and followed by a simple online separation on a HILIC guard column.
Fig. 3. Chromatograms of sample test solutions and standards, developed using water on a cellulose plate (a) and n-propanol–water–acetic acid (4:2:1, v/v/v) on a cellulose plate (b). Plates were derivatised using DMACA detection reagent. The tracks are juniper seed sample test solution, application volume 10 L (1), (-)-epicatechin (2), (+)-catechin (3), procyanidin B2 (4), (-)-epigallocatechin (5) and pomegranate sample test solution, application volume 5 L (6).
S. Smrke, I. Vovk / J. Chromatogr. A 1289 (2013) 119–126
123
Fig. 4. TLC–MS spectrum from a juniper seed extract of (a) co-eluting proanthocyanidin dimers (m/z 577) and trimers (m/z 865) eluted from a cellulose plate, developed with water, sampled at RF 0.40 and (b) a spectrum of flavonoids co-eluting with proanthocyanidin dimers and trimers eluted from a cellulose plate, developed with n-propanol–water–acetic acid (4:2:1, v/v/v), sampled at RF 0.71.
after the elution head, before the eluent entering the ion source for a very simple online separation. Methanol proved to be a too strong eluent on HILIC to accomplish any gain in signal, so acetonitrile was added to the eluent. Methanol–acetonitrile 1:1, 1:2 and 1:3 (v/v) were tested and 1:2 gained the best results (Fig. 2b), showing no signals of background peaks as observed when eluting without online separation (Fig. 2a). Unfortunately this approach was shown to work only on monomers, because they interact very little with HILIC stationary phase, as there was no MS signal observed when analysing procyanidin B2. This is probably due to procyanidin B2 having a far higher retention on the HILIC stationary phase, so it was not possible to separate it from interfering compounds. 3.4. TLC–MS of sample test solutions As described earlier, it was only possible to detect monomers on silica plates by means of TLC–MS and the use of these plates was therefore limited to this application. Cellulose HPTLC plates were far more interesting for use with TLC–MS as two different developing solvents, pure water and n-propanol–water–acetic acid (4:2:1, v/v/v), offer different separation performance. Methanol at a flow rate of 0.2 mL/min was used as the eluent and no additional HILIC separation was needed. If screening for only specific flavanols with known fragmentation pattern was needed, the preferred detection
Fig. 5. TLC–MS spectrum of pomegranate fruit peel sample test solution, showing proanthocyanidin dimers composed of two (+)-catechin units (m/z 577), one (+)catechin and one (+)-gallocatechin units (m/z 593) and two (+)-gallocatechin units (m/z 609).
Fig. 6. TLC × MS chromatograms of a juniper seed extract test solution (a) developed on cellulose HPTLC plates using water and (b) on n-propanol–water–acetic acid (4:2:1, v/v/v). Application volumes were 10 L.
124
S. Smrke, I. Vovk / J. Chromatogr. A 1289 (2013) 119–126
method would be SRM. Not all analytes present in analysed samples are available as standards to optimise their fragmentation pattern; therefore to broaden the range of detectable compounds only full scan MS detection was used. Because of lack of resolution on HPTLC plates, compared to HPLC, high selectivity of DMACA is welcome in case of co-elution with non-flavanols, but any overlap of peaks cannot be detected by a single developing solvent. A combination of different developing solvents [27] is needed to separate the most common monomers and proanthocyanidins. Although MS cannot distinguish between flavanol epimers, it can give information about degree of polymerisation of proanthocyanidins and in some cases the type of monomer unit bonding or types of monomers present. Sample test solution was developed using cellulose sorbent, with both developing solvents and derivatised using the DMACA detection reagent. When crude extracts were applied to the plates, interfering compounds of similar chemistry and abundant in high quantities, e.g. ellagic acid and ellagitannins [37] in pomegranate fruit, overloaded the stationary phase and produced distorted chromatograms. Furthermore, when crude extracts were applied to the HPTLC plates, it was possible by means of TLC–MS to detect only the most abundant flavanols (+)-catechin from juniper seeds and (+)-catechin and (+)-gallocatechin from pomegranate peel. Consequently, a sample preparation step was necessary and used for all analyses presented here. Chromatograms of sample test solutions (application volume: juniper seed sample 10 L, pomegranate fruit peel sample 5 L) were compared with those of (+)-catechin, (−)-epicatechin, procyanidin B2 and (−)-epigallocatechin (application mass 100 ng) (Fig. 3). The cellulose plates provide information about the epimers present in samples. Both developing solvents that were used (Fig. 3a and b) separate (+)-catechin from (−)-epicatechin. When water was used as a developing solvent (Fig. 3a) the most abundant peak detected in both samples was (+)-catechin and compared to (+)-catechin, only very small amounts of (−)-epicatechin were detected. A peak that could not be identified solely by TLC caused interference with (−)-epicatechin when n-propanol–water–acetic acid (4:2:1, v/v/v) was used as the developing solvent. The most intense peak of the juniper seed sample seen in the chromatogram developed on cellulose plate using water (Fig. 3a) was probed with TLC–MS. Co-elution of proanthocyanidin dimers ([M−H]− , m/z 577) and trimers ([M−H]− , m/z 865) with the main peak of (+)-catechin ([M−H]− m/z 289) was observed. All three flavanols were detected for the first time in juniper seeds. We were unable by means of TLC–MS to detect a monomer peak (m/z 289) when sampling at the RF position of the (−)-epicatechin standard. Consequently, we confirmed that (−)-epicatechin could be present only in very small quantities and that the peaks at m/z 577 and m/z 865 represent oligomers of (+)-catechin monomeric units (Fig. 4a). Since no (−)-epicatechin was found on the cellulose plate developed with water, the interfering peak seen on the cellulose plates developed with n-propanol–water–acetic acid (4:2:1, v/v/v) was tested (Fig. 3b). No monomers could be detected, but proanthocyanidin dimers and trimers were detected together with other flavonoids (Fig. 4b), such as quercetin (m/z 301) [43]. Proanthocyanidins in pomegranate fruit are composed of (+)-catechin and (+)-gallocatechin monomeric units [35]. The additional hydroxyl group in (+)-gallocatechin enables the distinction of these monomeric units by MS. By sampling the peak seen on cellulose plates developed with n-propanol–water–acetic acid (4:2:1, v/v/v) at the position of (−)-epicatechin standard (Fig. 3b) (+)-gallocatechin was detected ([M−H]− , m/z 305); there was no peak at m/z 289. The presence of (−)-epigallocatechin was excluded because both epimers are separated in the chromatogram (Fig. 3b). Dimers composed only of (+)-catechin ([M−H]− , m/z 577),
(+)-gallocatechin ([M−H]− , m/z 609) or mixed dimers ([M−H]− , m/z 593) could also be identified from the MS (Fig. 5). We attempted to detect proanthocyanidin trimers but were unsuccessful, superior separation capacity or sample preparation would be necessary.
Fig. 7. TLC × MS chromatograms of pomegranate fruit peel test solutions developed on cellulose HPTLC plates using water (a) or n-propanol–water–acetic acid (4:2:1, v/v/v) as developing solvent (b, c). Application volumes: 5 L (a, b), 20 L (c).
S. Smrke, I. Vovk / J. Chromatogr. A 1289 (2013) 119–126
3.5. Comprehensive TLC × MS of sample test solutions To detect analytes using TLC–MS, when their position on the plate is unknown, the application termed “comprehensive TLC × MS” is used. It is not possible to achieve such detection by sampling a chromatogram of only one application on the plate. There are two reasons for this; first, after the elution is complete and the elution valve is closed, there is still a significant pressure from the remaining mobile phase in the elution head. When the head is lifted from the plate, a small amount of mobile phase usually spills on the plate, wetting the stationary phase near the sampling position. Because the mobile phase must be a very strong eluent the chromatogram around the sampling position becomes distorted. Second, lifting the elution head from the plate frequently damages the adjacent stationary phase. Attempts to sample TLC–MS from previous elutions nearby proved to be unsuccessful because the damage to the stationary phase caused eluent to leak from the elution head. Based on the information obtained by sampling peaks as seen in the chromatograms derivatised with DMACA, we performed comprehensive TLC × MS as follows. Juniper seed sample test solutions (10 L) were developed on cellulose plates with both developing solvents and TLC × MS was performed between RF 0.19 and 0.60 in RF steps of 0.032 (Fig. 6a) when water was used as the eluent and between RF 0.61 and 0.84 in RF steps of 0.029 (Fig. 6b) when n-propanol–water–acetic acid (4:2:1, v/v/v) was used as the developing solvent. Overlapping peaks of (+)-catechin, proanthocyanidin dimers and trimers could be seen when the plates were developed with water. When a TLC × MS chromatogram (Fig. 6a) is compared to an image of the HPTLC plate derivatised with DMACA detection reagent (Fig. 3a, track 1), a very large difference in sensitivity between (+)-catechin and proanthocyanidin oligomers is observed. The oligomers could only barely be detected on the derivatised plate if the image is computer enhanced, but with the TLC × MS the signal height is approximately the same. Peak broadening detected by comprehensive TLC × MS might have occurred as a result of undersampling and deviation between two chromatograms where TLC × MS is performed, but the effect is insufficient to produce peak overlap. Sample test solutions of pomegranate peel extract developed on cellulose plates were probed by TLC × MS at RF between 0.34 and 0.66 in RF steps of 0.029 (Fig. 7a) when developed with water (application volume 5 L). Two different application volumes were used when developed with n-propanol–water–acetic acid (4:2:1, v/v/v), 5 L (RF between 0.57 and 0.93 in steps of 0.033, Fig. 7b) and 20 L (RF between 0.57 and 0.93 in steps of 0.033, Fig. 7c). When water was used as a developing solvent, separation but not baseline separation, between compounds was achieved. Use of n-propanol–water–acetic acid (4:2:1, v/v/v) as a developing solvent gave a very good separation between (+)-catechin and (+)-gallocatechin by TLC × MS, but no other flavanol peaks could be detected by means of TLC–MS. When a larger quantity of sample was applied to the HPTLC cellulose plate both monomers were still separated and dimers could be easily detected.
Alternatively, online separation must be employed to separate any possible ionisation inhibitors from the analytes, therefore for flavanol analysis we favoured the use of cellulose sorbent where sensitivity was the highest. Comprehensive TLC × MS offers a better view of the analysed TLC chromatogram and in addition to TLC–MS can provide information about the retention behaviour of analytes on TLC plates. For analysis of flavanols by TLC–MS or TLC × MS, separation on cellulose sorbent, developed with npropanol–water–acetic acid (4:2:1, v/v/v) or pure water, eluted with methanol at a flow rate of 0.2 mL/min without additional HILIC separation after elution and full scan MS detection is the method of choice. Acknowledgement This study was carried out with financial support from the Slovenian Research Agency and EN-FIST Centre of Excellence (Dunajska 156, SI-1000 Ljubljana, Slovenia). References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16]
[17] [18] [19] [20] [21]
[22]
[23] [24] [25] [26]
4. Conclusions We have shown that TLC–MS is a very versatile tool for identification and detection of compounds on TLC plates. It is simple and provides valuable information about analytes even if there is a very specific and sensitive derivatisation reagent available, as is the case with flavanols. If there are problems with TLC–MS sensitivity when analytes are eluted from silica gel sorbents, the SRM mode can be applied to detect a given compound.
125
[27] [28] [29] [30] [31] [32] [33] [34] [35]
S.C. Cheng, M.Z. Huang, J. Shiea, J. Chromatogr. A 1218 (2011) 2700. G. Morlock, W. Schwack, J. Chromatogr. A 1217 (2010) 6600. G. Morlock, W. Schwack, Trends Anal. Chem. 29 (2010) 1157. S.P. Pasilis, G.J. Van Berkel, J. Chromatogr. A 1217 (2010) 3955. ´ B. Simonovska, A. Albreht, D. Agbaba, J. Chromatogr. A 1218 I. Vovk, G. Popovic, (2011) 3089. H. Luftmann, Anal. Bioanal. Chem. 378 (2004) 964. ´ D. Staszek, M. Waksmundzka-Hajnos, T. KowalM. Sajewicz, Ł. Wojtal, M. Natic, ska, J. Liq. Chromatogr. Relat. Technol. 34 (2011) 848. ´ Ł. Wojtal, M. Waksmundzka-Hajnos, T. KowalM. Sajewicz, D. Staszek, M. Natic, ska, J. Liq. Chromatogr. Relat. Technol. 34 (2011) 864. M. Sajewicz, Ł. Wojtal, M. Hajnos, M. Waksmundzka-Hajnos, T. Kowalska, J. Planar Chromatogr. 23 (2010) 270. M. Sajewicz, D. Staszek, M. Natic, M. Waksmundzka-Hajnos, T. Kowalska, J. Chromatogr. Sci. 49 (2011) 560. ´ B. Simonovska, A. Albreht, I. Vovk, J. Chromatogr. A 1231 (2012) Z. Rodic, 59. V. Glavnik, B. Simonovska, A. Albreht, I. Vovk, J. Planar Chromatogr. 25 (2012) 251. K. Naumoska, B. Simonovska, A. Albreht, I. Vovk, J. Planar Chromatogr. 26 (2013) 125. P. Abu-Rabie, N. Spooner, Anal. Chem. 81 (2009) 10275. K. Heinig, T. Wirz, A. Gajate-Perez, Bioanalysis 2 (2010) 1873. ˇ V. Koleˇckáˇr, Z. Reháková, E. Brojerová, K. Kuˇca, D. Jun, K. Macáková, L. Opletal, P. Draˇsar, L. Jahodáˇr, J. Chlebek, L. Cahlíková, Chem. Listy 106 (2012) 113. I. Spranger, B. Sun, A.M. Mateus, V. de Freitas, J.M. Ricardo-da-Silva, Food Chem. 108 (2008) 519. T. Kurosawa, F. Itoh, A. Nozaki, Y. Nakano, S. Katsuda, N. Osakabe, H. Tsubone, K. Kondo, H. Itakura, Atherosclerosis 179 (2005) 237. T.P. Kenny, C.L. Keen, P. Jones, H.J. Kung, H.H. Schmitz, M.E. Gershwin, Exp. Biol. Med. 229 (2004) 255. M. Tomaru, H. Takano, N. Osakabe, A. Yasuda, K. Inoue, R. Yanagisawa, T. Ohwatari, H. Uematsu, Nutrition 23 (2007) 351. D. Ramljak, L.J. Romanczyk, L.J. Metheny-Barlow, N. Thompson, V. Knezevic, M. Galperin, A. Ramesh, R.B. Dickson, Mol. Cancer Ther. 4 (2005) 537. H.A. Weber, A.E. Hodges, J.R. Guthrie, B.M. O‘Brien, D. Robaugh, A.P. Clark, R.K. Harris, J.W. Algaier, C.S. Smith, J. Agric. Food Chem. 55 (2007) 148. J. Valls, S. Millán, M.P. Martí, E. Borràs, L. Arola, J. Chromatogr. A 1216 (2009) 7143. I. Fecka, W. Cisowski, M. Luczkiewicz, J. Planar Chromatogr. 14 (2001) 405. I. Vovk, B. Simonovska, P. Vuorela, H. Vuorela, J. Planar Chromatogr. 15 (2002) 433. I. Vovk, B. Simonovska, S. Andrenˇsek, H. Vuorela, P. Vuorela, J. Chromatogr. A 991 (2003) 267. I. Vovk, B. Simonovska, H. Vuorela, J. Chromatogr. A 1077 (2005) 188. R. Nowak, J. Planar Chromatogr. 20 (2007) 43. E. Reich, A. Schibli, V. Widmer, R. Jorns, E. Wolfram, A. DeBatt, J. Liq. Chromatogr. Rel. Technol. 29 (2006) 2141. T. Hofmann, L. Albert, T. Rétfalvi, J. Planar Chromatogr. 17 (2004) 350. L. Pobłocka-Olech, M. Krauze-Baranowska, M. Wiwart, J. Planar Chromatogr. 20 (2007) 61. V. Glavnik, B. Simonovska, I. Vovk, J. Chromatogr. A 1216 (2009) 4485. V. Glavnik, B. Simonovska, I. Vovk, J. Planar Chromatogr. 23 (2010) 230. ´ D. Aˇsperger, S. Babic, ´ V. Glavnik, B. Simonovska, I. Vovk, D. Mutavdˇzic´ Pavlovic, J. Planar Chromatogr. 24 (2011) 482. C. Ben Nasr, N. Ayed, M. Metche, Z. Lebens. Unters. Forsch. 203 (1996) 374.
126
S. Smrke, I. Vovk / J. Chromatogr. A 1289 (2013) 119–126
[36] N. Miceli, A. Trovato, P. Dugo, F. Cacciola, P. Donato, A. Marino, V. Bellinghieri, T.M. La Barbera, A. Güvenc¸, M.F. Taviano, J. Agric. Food Chem. 57 (2009) 6570. [37] N.P. Seeram, L.S. Adams, S.M. Henning, Y. Niu, Y. Zhang, M.G. Nair, D. Heber, J. Nutr. Biochem. 16 (2005) 360. [38] G.W. Plumb, S. de Pascual-Teresa, C. Santos-Buelga, J.C. Rivas-Gonzalo, G. Williamson, Redox Rep. 7 (2002) 41. [39] S. de Pascual-Teresa, C. Santos-Buelga, J.C. Rivas-Gonzalo, J. Agric. Food Chem. 48 (2000) 5331.
[40] S. de Pascual-Teresa, J.C. Rivas-Gonzalo, C. Santos-Buelga, Int. J. Food Sci. Technol. 35 (2000) 33. [41] R. Engelshowe, Planta Med. 49 (1983) 170. [42] F. Martz, R. Peltola, S. Fontanay, R.E. Duval, R. Julkunen-Tiitto, S. Stark, J. Agric. Food Chem. 57 (2009) 9575. [43] M. Innocenti, M. Michelozzi, C. Giaccherini, F. Ieri, F.F. Vincieri, N. Mulinacci, J. Agric. Food Chem. 55 (2007) 6596. [44] A.G.H. Lea, J. Sci. Food Agric. 29 (1978) 471.