Advertisement

Multidimensional Separation of Natural Products Using Liquid Chromatography Coupled to Hadamard Transform Ion Mobility Mass Spectrometry

  • Wenjie Liu
  • Xing Zhang
  • Richard Knochenmuss
  • William F. Siems
  • Herbert H. HillJr.Email author
Research Article

Abstract

A high performance liquid chromatograph (HPLC)was interfaced to an atmospheric drift tube ion mobility time of flight mass spectrometry. The power of multidimensional separation was demonstrated using chili pepper extracts. The ambient pressure drift tube ion mobility provided high resolving powers up to 166 for the HPLC eluent. With implementation of Hadamard transform (HT), the duty cycle for the ion mobility drift tube was increased from less than 1% to 50%, and the ion transmission efficiency was improved by over 200 times compared with pulsed mode, improving signal to noise ratio 10 times. HT ion mobility and TOF mass spectrometry provide an additional dimension of separation for complex samples without increasing the analysis time compared with conventional HPLC.

Graphical Abstract

Keywords

HPLC IMS TOF-MS ESI Hadamard transform Chili pepper 

Introduction

Natural products from plants and other living organisms often have pharmacologic or biological activities for use in pharmaceutical drug discovery and design. Screening and identification of bioactive natural products have emerged as a significant challenge to separation science because of their inherent complexity and variability [1, 2, 3]. Considering that most natural products are nonvolatile compounds, such as alkaloids and flavonoids, liquid chromatography is the method of choice for their separation [4, 5]. For example, chili pepper extract has been applied to treat type-I diabetes, and the major bioactive component from chili pepper was found to have anticarcinogenic potential [6, 7]. Capsaicin(8-methyl-N-vanillyl-6-nonenamide) is the major irritant compound, producing a burning sensation in any tissue with which it comes into contact [8]. Capsaicin and several related compounds are called capsaicinoids and are produced as secondary metabolites by chili peppers. Current analytical methods of characterizing chili pepper extract include high performance liquid chromatography (HPLC) [9, 10], LC-MS [11, 12], and electrophoresis [13, 14]. However, the number of compounds identified from chili pepper is limited by the resolving power of one-dimensional HPLC methods.

Ion mobility spectrometry (IMS) has been recognized as an important analytical method for its speed, sensitivity, and unique separation mechanism [15, 16, 17]. During the past decade, ion mobility has expanded its application from chemical warfare agent, illicit drug, and explosive detection to pharmaceutical analysis, and food safety inspection [18, 19, 20]. Different ion mobility instruments, such as drift tube ion mobility spectrometers and differential ion mobility spectrometers (field asymmetric ion mobility spectrometer) were developed for these various applications. Hyphenated with mass spectrometry, especially time of flight mass spectrometry, ion mobility mass spectrometry provided a multidimensional technique for complex sample analysis such as those required for metabolomics and proteomics research. Ion mobility provides separation complementary to mass/charge separation of mass spectrometry [21, 22]. Generally, IMMS can provide several times the peak capacity of mass spectrometry alone. More importantly, mobility–mass relationships provide added information in the form of mobility–mass trend line data, which helps to identify unknown compounds of similar structures.

Two-dimensional liquid chromatography was developed to resolve complex samples [23, 24, 25]. Although 2D liquid chromatography improved separations of complex samples, it suffers from low resolution and time-consuming slow separation steps. In some cases, there is a mismatch of solvents between the first and second LC dimension [26, 27, 28]. Recently, another comprehensive 2D separation technique based on liquid chromatography and ion mobility was introduced to separate complex samples [29, 30, 31, 32, 33]. Based on traveling wave ion mobility spectrometry (TWIMS), a relatively new ion mobility technology that provides an extra separation dimension to mass spectrometry, HPLC-IMS-MS, has been successfully applied to metabolite profiling, proteomics, the determination of free desmosine and isodesmosine as urinary biomarkers of lung disorder, and the determination of hydroxylated polybrominated diphenyl ethers. HPLC combined with ion mobility/time-of-flight (TOF) mass spectrometry was also used to characterize a combinatorial peptide library containing 4000 peptides, demonstrating that mobility separation between the HPLC and TOF measurement dimensions makes it possible to resolve many peptide isomers that have identical retention times and masses [33].

One important aspect of ion mobility spectrometry when coupled with chromatography and mass spectrometry is resolving power. For typical operating conditions of 150 °C, 20 cm drift length, and 500 V/cm, ambient pressure drift tube instruments have conditional resolving powers as high as 160 using a gate pulse of 100 us, providing reliable separation of isomers and isobars within tens to hundreds of milliseconds [34, 35]. Another advantage of drift tube ion mobility is the direct calculation of collision cross sections. Other mobility methods, such as traveling wave ion mobility spectrometry, can also estimate cross section using nonlinear calibration methods [36, 37].

One of the major disadvantages of drift tube ion mobility spectrometry is an inherent lack of sensitivity due to its low duty cycle (~1%). The ion gate admits a discrete packet of ions into the drift region. To avoid spectral overlap, the drift time measurement of each ion packet must be completed prior to injection of the next ion packet. Generally, less than 1% of total ions from the ion source are utilized for separation. One method to compensate for this low duty cycle is summing the drift time and flight time datasets with multiple runs at the expense of time, especially for HPLC experiments [24].

Another approach to improve the duty cycle is the use of an ion trap as the ion gate [38, 39]. With this approach, ions are accumulated during the separation cycle of the ion mobility spectrum and then injected into the spectrometer as a pulse of ions. In theory, the duty cycle can be improved with this method to almost 100%. In practice, however, charge repulsion and ion diffusion reduce the efficiency. Although ion-trapping experiments can improve the duty cycle of an IMS coupled to an MS, they must be operated at low pressure, reducing the resolving power of the IMS and the accuracy of collision cross section measurements, while increasing the level of instrumental complexity and expense.

Hadamard transform ion mobility mass spectrometry applies pseudorandom sequences to control the opening of ion gate and regenerate the ion mobility spectrum via deconvolution [40, 41]. In theory, the use of Hadamard multiplexing increases the duty cycle of the instrument to 50%, so the limit of detection (LOD) and signal to noise are significantly improved. In the absence of any other effects, the improvement of signal to noise is given by Equation 1:
$$ Enhancement=\sqrt{\frac{0.5}{D{C}_p}} $$
(1)

DC p : the duty cycle of conventional pulsed mode

The Hadamard approach using IMS with MS detection has demonstrated improvement in sensitivity without losing resolving power. Evaluation on its capability of rapid analysis for complex mixtures was achieved using human plasma. Approximately 250 metabolite ions were detected within 2 min with high resolving power (≥100) [42, 43, 44].

This paper investigates for the first time the addition of an HT IMS between liquid chromatographic separation and mass spectrometry. When standard drift tube IMS is added between the LC and MS, sensitivity is not sufficient at the speed required to detect individual LC peaks. Addition of the HT-IMS provides a rapid scanning device that is sufficiently sensitive to detect individual compounds eluting from the liquid chronograph. This third mode of separation is expected to enhance resolving power over that possible with of HPLC-MS alone, producing improved peak capacity for the separation, detection, and identification of individual compounds in a complex sample. While three-dimensional separations have been reported previously with 2D HPLC-MS, advantages of using IMS as the second dimension rather than another LC are speed (real time separation) and enhanced resolving power. Full HT-IMS scans require only about 0.1-0.2 s, so the LC remains the speed-limiting factor. This is also sufficiently fast to retain all of the information from the separation of the first LC dimension. The performance of this HPLC×HTIMS-TOFMS instrumental combination was evaluated using a chili pepper extract for sensitivity, peak capacity, and orthogonal separation between the first HPLC dimension and the second IMS dimension.

Experimental

Chemicals and Reagents

All HPLC grade solvents were obtained from Sigma-Aldrich (Buchs, Switzerland) and used without further purification. Glacial acid was purchased from J. T. Baker (Phillipsburg, NJ, USA).

Chromatography

A MicroTech Ultra Plus II high performance liquid chromatography with dual solvent pump high-pressure gradient system was used in this study. A Rheodyne 20-μL auto injector was used for sample injection. Samples were separated on a 100 mm * 2.1 mm, 5-μm particle, Waters Atlantis RP C18 column (Waters, Milford, MA, USA). A T piece was used as post-column splitter and the split ratio was adjusted to 50:1.

The chromatographic elution was conducted with binary mobile phase gradient consisting of water (A) and methanol (B). Initial gradient conditions were set to 50% B at the flow rate of 0.2 mL/min before incorporating a linear gradient increasing to 100% B over 10 min and held for 10 min. At 21 min, the gradient was returned to the initial conditions and equilibrium for 5 min. The column temperature was maintained at room temperature.

Ion Mobility Spectrometry

An ambient pressure stacked ring drift tube IMS similar to the device used in this system has previously been described [34, 45]. The IMS tube consisted of two regions: an 8-cm desolvation region and a 23-cm drift region separated by a Bradbury Nielsen ion gate made with gold-plated tungsten wires spaced 0.5 mm apart. The ion gate was operated with a closure voltage of ±38 V and an electrical gate width of 150 μs in pulsed mode and 180 μs chip width in Hadamard mode. The electric field across the IMS drift region was 413 V/cm. The IMS tube was maintained at 200 °C during the whole experiment. Atmospheric pressure was 93.3 kPa during the experiment process with clean air used as drift gas (900 mL.min–1). A 150 μm inner diameter silica capillary tube with polyimide coating was used to conduct the eluent from HPLC column to a T-piece and a 46 μm inner diameter silica capillary tube was used as ESI needle. A 3.0 kV bias potential was applied to the ESI needle relative to the drift voltage of the drift tube (Figure 1).
Figure 1

Schematic of HPLC-HTIMS-TOF instrument used for multidimensional separation. The HPLC phase separation is carried out with a MicroTech Ultra Plus II HPLC (not using the detector). Positive electrospray ionization was used as sample introduction from HPLC to ion mobility drift tube with a T-piece as splitter affording a flow rate of 4 μL.min–1. TOF mass spectrometer worked with a V mode (sensitivity mode)

Time of Flight Mass Spectrometry

The time of flight mass spectrometer was purchased from TOFWERK AG (Thun, Switzerland) and has been described elsewhere [32]. Briefly, the interface between the IMS and the TOF region of the MS consisted of a pinhole nozzle with a 300 μm diameter. The pressure inside the interface was stepped down in two stages, from atmospheric pressure (approximately 950 mbar) to 2–4 mbar within the interface. Two segmented quadrupole ion guides guided the ions traversing the pressure interface region towards the ion focusing lenses.

Sample Preparation

Whole dry hot chili pepper was purchased from a local grocery store and crushed by hand. Ten grams of chili pepper was manually ground with a mortar and pestle; 0.5 g of chili pepper powder was weighed into a 20 mL glass vial and extracted with 10 mL of 80/20% methanol:water. The mixture was sonicated for 30 min and 1.5 mL of mixture was centrifuged under 12,000 rpm for 10 min. The supernatant was filtrated with 0.45 μm membrane and used without further preparation for HPLC analysis.

Data Analysis

Data was acquired with the Tofwerk software and imported to Igor Pro 6.31; retention time of chromatogram and drift time of ion mobility spectrum were obtained with TOFDAQviewer. Two dimensional HPLC-IMS and IMMS spectra were obtained with IMSviewer (Tofwerk, Thun, Switzerland).

The resolving powers of each compound were calculated with the drift time (td) of the ion divided by the temporal peak width-at-half-height (w0.5) of the ion mobility peak. This relationship is given by Equation 2:
$$ {R}_m=\frac{t_d}{w_{0.5}} $$
(2)
where R m is the measured resolving power.

Results and Discussion

IMS-LC Separation

Figure 2a provides a typical example of two-dimensional HPLC-IMS data from a chili pepper extract. The horizontal axis denotes the chromatographic retention time in seconds whereas the vertical axis is the ion mobility drift time in ms. The total ion chromatogram shown in Figure 2b is the separation profile of the HPLC separation demonstrating a peak capacity of 59.2. Ion mobility drift times ranged from 15 to 60 ms. Each mass spectrum corresponds to a specific retention and drift time. The mass resolution used in this study varied from about 5000 to 6500. With a mass resolution of 5000, it was possible to obtain the chemical formula from the exact mass to charge of ions. Each retention time point provided an IMMS spectrum.
Figure 2

(a) A two-dimensional separation by HPLC and ion mobility spectrometry from chili pepper extract in the Hadamard transform mode; mass range is from 30 to 800 m/z. The horizontal axis is the HPLC retention time; the vertical axis is the ion mobility drift time. (b) The total ion chromatogram (TIC). (c) The total ion mobility spectrum (TIMS). (d) A mobility selected chromatogram having a drift time 36.96 ms. (e) A selected ion mobility spectrum having a retention time of 862 s. (f) A mobility (drift time 36.96) and retention time (862 s) selected mass spectrum correspond to the lines on the 2D plot

Several observations about the LC-IMS separation can be observed. First, the two separations are orthogonal to one another. That is, there is no obvious correlation between mobility and retention. In some cases, compounds co-elute through the chromatogram and are separated by IMS and in other cases ions have the same mobility but are separated in the chromatographic space. Several of the compounds are not well separated by this chromatographic condition and “smear” (elute continuously) throughout the chromatogram. This is observed in the high background response in Figure 5b and by the several response “lines” that show up in the 2D spectrum. These ions are well resolved in the IMS spectrum forming sharp peaks at specific drift times. A second observation is that IMS separates the mixture with a higher resolving power than the liquid chromatographic system. In this case, the average resolving power with post-processing for IMS was 91, which corresponds to a number of theoretical plates of N = 5.55R2 = 46,000, significantly higher than most LC separations. Finally, there are a number of compounds that eluted with the void volume in the LC that can be separated nicely with IMS. Three of the target compounds (capsaicin, dihydrocapsaicin, and capsiamide) in chili peppers were identified with standards and are labeled on the 2D spectrum. Dihydrocapsaicin and capsiamide show similar but not identical drift times, and were separated well with chromatography. Capsaicin and dihydrocapsaicin were not well separated by chromatography but completely separated in the ion mobility mode.

Hadamard Transform Advantage

Traditional drift tube ion mobility mass spectrometry (DT IMMS) affords a rapid gas-phase separation step before MS analysis, but the standard pulsed operation of the IMS limits the application of DT IMMS to HPLC. Long gate pulses decrease the resolving power of ion mobility peaks, whereas short gate pulses decrease the quantity of ions introduced into the drift tube and decrease the sensitivity consequently. In our experiments, sensitivity drops significantly when the gate pulse is shorter than 150 us, so we used 150 us as the gate pulse width. Since the whole drift time range in the pulsed mode experiment was set to 60 ms and the gate pulse width was 150 us, the duty cycle (DC) of pulsed mode ion mobility spectrometry was
$$ D{C}_p=\frac{150\mu s}{60\times 1000\mu s}=0.0025 $$
(3)
Thus, only 0.25% ion transmission efficiency was obtained in pulsed mode, at least 99.75% of the ions were lost at the ion gate. For HT mode, the DC was fixed to 0.5, so the ion transmission efficiency was improved approximately 200 times, which could result 200 times improvement of all mass intensities. Figure 3 shows the intensity differences between the two modes. For HPLC separation, this was observed with the peak height of the same mass selected chromatogram at the same retention time. In pulsed mode, the intensity for the selected ion chromatogram peak at m/z 270 was 0.0026 arbitrary units (AU), whereas the peak intensity was 0.5094 AU in HT mode. The ion intensity for the m/z 270 peak was increased 197-fold, which was very close to the theoretical expectation.
Figure 3

Three-dimensional IMMS plot of chili pepper extract separation illustrating the intensity difference in pulsed mode compared to HT mode. (a) IMMS was operated in pulsed mode with pulse width of 150 μs, with a duty cycle of 0.25%. (b) IMMS was operated under HT mode, with a duty cycle of 50%. Both (a) and (b) used the average ion intensity at each chromatographic retention time

For the ion mobility spectrometry, a similar enhancement factor was observed. The ion mobility spectrum at LC retention time 508 s and at m/z 306 showed a mobility of 35.52 ms with intensity of 114 in HT mode and 0.6 in pulsed mode. The intensity in HT mode was 194 times that of the pulsed mode. The comparative results are shown in the Supplementary Material, Figure S1.

Previous reports showed no significant change of resolving power in Hadamard mode compared with pulsed mode [46, 47]; here we compare the 15 major mobility peaks. Though we used 20% wider gating pulse in HT mode, the raw resolving power was not less than in pulsed mode. The average resolving power peak in pulsed mode was 72.8, whereas the average resolving power in HT mode was 71.5. These results indicate that HT slightly improved the raw resolving power. Furthermore, the resolving power of HT spectra improved significantly using post-processing methods [48]. For all 15 peaks, the resolving powers are much higher for the HT mode with sharpening. Table 1 compares resolving powers between the pulse and Hadamard modes for each of the 15 peaks.
Table 1

Resolving Power Comparison for Pulsed and Hadamard Modes

Peak list

m/z

Retention time

Drift time

R m (P)a

R m (HT)

R m (HT)b

1

82.08

144

17.35

50.2

52.0

105

2

137.11

504

35.29

43.8

53.06

83

3

137.11

549

36.75

82.6

88.1

106

4

146.12

109

22.21

70.2

65.8

134

5

152.12

549

22.70

90.7

88.1

135

6

153.11

522

22.66

78.2

73.7

125

7

170.19

512

25.53

66.9

66.2

127

8

181.18

342

25.46

85.7

73.1

126

9

219.10

90

25.48

66.1

51.2

102

10

270.41

846

37.13

87.6

85.8

166

11

279.28

693

33.72

47.5

61.2

119

12

294.34

495

35.49

69.3

78.4

158

13

306.34

510

35.32

64.8

62.7

138

14

308.36

549

36.80

67.8

77.6

163

15

328.34

504

35.41

75.4

95.3

154

Average

72.8

71.5

129.4

aMeasured resolving power in pulsed mode.

bmeasured resolving power in Hadamard transform mode with denoise 1 and sharpening 548.

All measured resolving power in both pulsed mode and HT mode are average values for mass selected chromatographic peaks. All drift times are from pulsed mode experiment and did not calibrate to exclude the drift time contribution outside of the drift tube.

The most important benefit of HT mode is the number of peaks distinguished because of the increase of sensitivity. In pulsed mode, only 28 peaks were detected, whereas more than 390 were found in Hadamard mode. Increased HT sensitivity resulted in approximately 10 times more peaks resolved by the combination of retention time and drift time.

Sampling rate of Hadamard transform ion mobility spectrometer for HPLC separation

Generally, to describe a whole HPLC peak, the sampling rate of the detector is required to be high enough for quantification and integration. In most cases, the area under a peak requires 10 points to describe it with some accuracy for an ideal chromatographic peak. In our experiment, we examined 10 selected ion chromatographic peaks with different retention time (Table 2), the average peak width at base was 35.1 s. In the Hadamard mode, we used a 184 ms measurement cycle, so that on average there were more than 190 points over one peak (5.43 Hz). If the chromatography peaks were sharper, the measurement cycle could be shorter; however, 190 points/peak proved to be adequate for both identification and quantification. By virtue of the sub-2 μm particle column and core-shell particle column, the column efficiency of liquid chromatography peaks improved rapidly during the past decades. Under optimum conditions, the peak width could be reduced to less than 10 s. Hadamard mode ion mobility spectrometry still provides an adequate sampling rate, and the measurement cycle time can be decreased. However, increased sampling rate increases the data dramatically; in this experiment, we used 20 cycles for a single stored data point, so there were still nearly 10 points over a single chromatographic peak, but the data size decreased 20 times.
Table 2

Typical HPLC Peak Width with Different Retention Time and Tailing Factor, Calculated from Selected Ion Chromatograms

Peak list

m/z

Retention time(s)

Peak width(s)

Tailing factor

1

216

118

35.7

1.10

2

238

126

37.4

1.16

3

240

640

30.6

1.22

4

240

847

28.9

1.07

5

245

894

34.0

1.12

6

254

744

32.3

1.06

7

294

493

42.6

1.23

8

306

508

40.9

1.25

9

295

1017

37.4

1.08

10

411

950

30.6

1.04

The Signal-to-Noise Advantage of IMS

Figure 5 shows an IMS-HPLC 2-D plot of the response of an ion at m/z 254. The x-axis is the mass-selected chromatogram and the y-axis is the mass-selected ion mobility spectrum. The spectrum shows two major peaks representing an isobaric pair (not identified). Figure 5c is the mass-selected chromatogram showing a signal-to-noise ratio of about 14. We summed up all ions with drift time between 15 to 60 ms and used a mass-selected and drift time-selected chromatogram. Intensity between 200 to 600 s was used as noise for the calculation of signal to noise ratio for both chromatograms. There was a 10-fold improvement of both mass-selected and drift time-selected chromatogram for the HPLC-IMS-TOF combination over the mass-selected chromatogram when no IMS was used (Figure 4). This can be attributed to the reduction of chemical noise by ion mobility spectrometry though the absolute intensity was reduced [49, 50, 51].
Figure 4

Signal to noise ratio improvement with drift time selected and mass selected chromatogram. (a) Mass selected (m/z 254) HPLC-IMS two-dimensional spectrum. Horizontal axis is retention time, vertical axis is drift time. (b) Both retention time (640–650 s) and ion (m/z 254) selected ion mobility spectrum. (c) Mass selected (m/z 254) chromatogram. (d) Only ion selected ion mobility spectrum of m/z 254. (e) Both mass selected (m/z 254) and drift time (32.90–33.10 ms) selected chromatogram afford S/N of 140.8

Two concerns about multidimensional separation with HPLC-IMS-MS are the limit of detection and method of quantification. Though a few methods have been proposed to integrate two-dimensional peaks in comprehensive two-dimensional spectrum [52, 53, 54], the most convenient way is to use only one-dimensional peaks. We compared the signal to noise of drift time and mass selected chromatogram, retention time, and mass selected ion mobility spectra. It was interesting to see that mass-selected chromatograms only afforded a S/N of 13.9, whereas drift time and mass-selected chromatogram gave a S/N of 140.8. From the mass-selected HPLC-IMS spectrum we clearly see the signal to noise difference in the retention time and drift time dimensions. This result suggested that drift time and mass selected chromatogram are more suitable for quantification and to calculate the limit of detection than retention time and mass-selected ion mobility spectra for lower noise level along the retention time dimension.

The Peak Capacity Advantage of IMS

As noted earlier, ion mobility is an orthogonal separation mechanism to reverse phase HPLC and, thus, the additional separation of IMS can expand the peak capacity of chromatography. Due to the molecular diversity of natural products, many similar structures and isomers exist. For many isomers, LC separation is difficult, and mass spectrometry cannot distinguish them. For example, for the selected ion chromatogram at m/z 295 shown in Figure 5, there were five isomeric peaks with different retention times. The third peak with a retention time of 685 s showed three clear ion mobility peaks at drift times of 33.56, 34.51, and 35.64 ms with resolving power of 101.3, 113.2.3, and 136.3. These three isomers could not be separated by chromatography but could be distinguished with ion mobility spectrometry.
Figure 5

Hadamard transform ion mobility separation for a single selected ion chromatographic peak. (a) Selected ion chromatogram of m/z 295. (b) Ion mobility spectra of selected ion chromatogram of m/z 295 at retention time of 686 s

Previous reports have demonstrated that ion mobility improves the peak capacity when added to mass spectrometry or liquid chromatography mass spectrometry. Even for very low resolution mobility separations such as FAIMS, the addition of mobility to LC-MS was reported to increase the peak capacity by a factor of eight [25].

For the calculation of ion mobility spectrometry peak capacity, previous reports define peak capacity in terms of the maximum number of peaks that could be resolved (using a 50% valley definition). Generally, the range over which peaks are separated divided by the full width at half maximum (FWHM) of selected peaks is used for calculation of peak capacity. However, this method over-estimates the separation power of ion mobility spectrometry. In our calculation, we used the following equation for a more accurate estimation of peak capacity(Pc):
$$ Pc=\frac{n\left({T}_2-{T}_1\right)}{{\displaystyle \sum_{i=1}^n{w}_{i,0.5}}} $$
(4)
where T 2 is the maximum drift time for selected mass range, T 1 is the minimum drift time for selected mass range. W i,o.5 is the peak width at half maximum, and n is the number of peaks for calculation. So, peak capacity in the mobility dimension is strongly dependent on the resolving power of ion mobility. Theoretically, the minimum W i,o.5 can be calculated from the following equation considering the initial gate pulse and ion diffusion. It is important to note that this applies only to pulsed mode. With post-processing, the minimum width in HT mode can be substantially less, by up to a factor of two.
$$ {w}_{0.5}=\sqrt{{w_g}^2+\left(\frac{16kT \ln (2)}{qV}\right){t_d}^2} $$
(5)
where w g is the initial gate pulse width, k is Boltzmann constant, T is the temperature in Kelvin, t d is the measured ion drift time, q is the electronic charge, and V is the voltage over the drift tube.
Under our ion mobility spectrometry conditions, the maximum drift time for selected mass range was 60 ms, the minimum drift time in spectra is 15.0 ms, the calculated W o.5 for 20, 30, 40, 50, and 60 ms peaks are 0.138, 0.207, 0.276, 0.345, and 0.414 ms. Thus, the maximum peak capacity of ambient pressure ion mobility spectrometry is estimated as:
$$ Pc=\frac{5\left(60-15\right)}{\left(0.138+0.207+0.276+0.345+0.414\right)}=163.0 $$

With post-processing, in HT mode, Pc can be up to 300.

We selected the IMMS frame at retention time 594 s to calculate the peak capacity of our ambient pressure HT ion mobility spectrometer as used in this study. Three peaks were used to determine an average peak width: m/z 152, 306, and 308 (the figure is provided in the Supplementary Material, Figure S2). The FWHM for three peaks were 0.262, 0.350, and 0.325 ms, and thus the estimated peak capacity for IMS was
$$ Pc=\frac{3\left(60-15\right)}{\left(0.262+0.350+0.325\right)}=144.1 $$
For the first dimension, the whole HPLC separation time was 1320 s and the average FWHM was 20.6 s, the effective separation time for HPLC was 1220 s not counting the dead time, so the peak capacity was:
$$ Pc=\frac{1220}{20.6}=59.2 $$

The HPLC-IMS two-dimensional separation peak capacity was 59.2 × 144.1 = 8530.7. This result is much higher than in the previous report, which estimated 900–1200 for the two-dimensional LC-IMS based on a low pressure drift tube ion mobility spectrometer [55]. It should be noted that post-processing provides up to two times improvements in width, and the peak capacity would improve correspondingly.

For liquid chromatography ion mobility TOF mass spectrometer separation, the resolution of our TOF instrument varies from 4500 to 6500; we conservatively estimate that the peak capacity of the mass spectrometer is nearly 5000, so the total peak capacity of this hyphenated HPLC-IMS-TOF mass spectrometer was estimated up to 4.27 × 107 without considering the correlation between drift time and m/z. Corrected peak capacity 1.75 × 107 is obtained by the separation space of IM-MS, which gives a 41% effective separation space from the total orthogonal space (Figure 6a).
Figure 6

IMMS 2D plot for trend analysis. (a) Averaged IMMS 2D plot from HPLC-IMMS analysis indicate a 41% separation space from the orthogonal. (b) Direct IMMS infusion analysis. (c) Amino acid mass mobility relationship plot. (d) Capsaicinoids mass mobility relationship plot. Line I represents saturated capsaisinoids, line II unsaturated capsaisinoids

Identification of Capsaicinoid Compounds from Chili Pepper Extracts

The identification of peaks from the multidimensional separation was performed manually. For a specific m/z, peaks in the selected ion chromatogram were chosen to obtain the ion mobility spectrum. In pulsed mode, all chromatogram peaks were selected to get the ion mobility spectrum; 30 peaks were distinguished in a 22 min LC run. For HT mode, the same method was used to get the LC-IMS peaks, but only intensities greater than 0.005 were selected to get the ion mobility spectrum. Thus, 162 peaks were distinguished in HT mode, nearly five times more than were observed in the pulsed mode. In this study, eight capsaicinoids were tentatively identified based on exact mass to charge ratio and their elution order from HPLC column (Table 3).
Table 3

Possible capsaicinoids and their drift time, retention time, and m/z

Mobility-Mass Trend Line

One attractive feature of ion mobility mass spectrometer analysis is the mobility–mass relationship analysis proved additional information for identification of compounds. In order to define the trend lines for peak identification, the IMMS 2D plot was averaged for all retention times and the results compared with those taken by direct infusion IMMS. The results of this comparison are shown in Figure 6. From these two plots, we can clearly differentiate similar features, though there were more peaks in the HPLC-IMMS plot than in the direct IMMS analysis. The addition of HPLC increases the sensitivity of the analysis over direct infusion method by reducing the charge competition of the matrix when all of the compounds are introduced into the electrospray ionization source simultaneously.

We selected amino acids and capsaicinoids to compare their trend line [58, 59]. Nineteen free amino acids were detected with HPLC-IMMS in mass- and retention time- selected ion mobility spectra. Because of the strong polarity differences of amino acids, their retentions times were from 81 to 162 s, and HPLC only affords a preliminary separation. Most of amino acids were resolved by ion mobility, and their drift times were from 16.91 to 27.17 ms. Capsaicinoids demonstrated two clear mobility mass linear relationship patterns; line I represents saturated capsaisinoids, and line II unsaturated capsaisinoids. Unsaturated capsaisinoids are denser than saturated capsaisinoids, so they travel faster in the drift tube and have smaller drift times. From this plot it is possible to predict capsaicinoid homologue drift times.

Conclusions

In this work, HPLC and a HT IMS-MS were coupled to improve both the capacity factor and the sensitivity of conventional pulsed mode IMS-MS. This HPLC-ESI-IMS-TOFMS was applied to the analysis of chili pepper extracts. The effective peak capacity in the ion mobility dimension was up to 141. The addition of HT to IMMS provided improved sensitivity without loss of resolution, with the intensity improvement nearly 200 times over that of the pulsed mode. In addition, the signal to noise ratio was improved 10 times in HT mode. HPLC-ESI-IMS-TOFMS provided an excellent separation for chili pepper extract with an overall peak capacity of 1.75 × 107. With the implementation of HT, the speed of ion mobility separation was improved sufficiently to track the HPLC peaks without losing the information of the chromatographic dimension. With the HT mode, liquid chromatography hyphenated with high resolving power ambient pressure drift tube ion mobility mass spectrometry is a feasible technique for the separation of natural product extracts. Eight capsaicinoid compounds were preliminarily identified with their exact mass to charge and elution order. In HPLC-IMMS mode, trend line analysis was still applicable for the mobility–mass relationship study. The most attractive character of the developed method is the comprehensive two-dimensional separation of complex samples without losing the resolution of the first dimension, with high resolving power of the second dimension. The addition of HPLC to IMMS provided improved sensitivity because of the decrease in charge competition of the components of the complex mixture when they are introduced after chromatographic separation rather than by direct infusion ESI.

Supplementary material

13361_2016_1346_MOESM1_ESM.docx (180 kb)
ESM 1 (DOCX 179 kb)

References

  1. 1.
    Smit, A.J.: Medicinal and pharmaceutical uses of seaweed natural products: a review. J. Appl. Phycol. 16, 245–262 (2004)CrossRefGoogle Scholar
  2. 2.
    Vuorela, P., Leinonen, M., Saikku, P., Tammela, P., Rauha, J.P., Wennberg, T., Vuorela, H.: Natural products in the process of finding new drug candidates. Curr. Med. Chem. 11, 1375–1389 (2004)CrossRefGoogle Scholar
  3. 3.
    Balunas, M.J., Kinghorn, A.D.: Drug discovery from medicinal plants. Life Sci. 78, 431–441 (2005)CrossRefGoogle Scholar
  4. 4.
    Eldridge, G.R., Vervoort, H.C., Lee, C.M., Cremin, P.A., Williams, C.T., Hart, S.M., Goering, M.G., O'Neil-Johnson, M., Zeng, L.: High-throughput method for the production and analysis of large natural product libraries for drug discovery. Anal. Chem. 74, 3963–3971 (2002)CrossRefGoogle Scholar
  5. 5.
    Koehn, F.E., Carter, G.T.: The evolving role of natural products in drug discovery. Nat. Rev. Drug Discov. 4, 206–220 (2005)CrossRefGoogle Scholar
  6. 6.
    Jun, H.-S., Park, T., Lee, C.K., Kang, M.K., Park, M.S., Il Kang, H., Surh, Y.-J., Kim, O.H.: Capsaicin induced apoptosis of B16-F10 melanoma cells through down-regulation of Bcl-2. Food Chem. Toxicol. 45, 708–715 (2007)CrossRefGoogle Scholar
  7. 7.
    Bley, K., Boorman, G., Mohammad, B., McKenzie, D., Babbar, S.: A comprehensive review of the carcinogenic and anticarcinogenic potential of capsaicin. Toxicol. Pathol. 40, 847–873 (2012)CrossRefGoogle Scholar
  8. 8.
    Cordell, G.A., Araujo, O.E.: Capsaicin - identification, nomenclature, and pharmacotherapy. Ann. Pharmacother. 27, 330–336 (1993)Google Scholar
  9. 9.
    Kurian, A.L., Starks, A.N.: HPLC analysis of capsaicinoids extracted from whole orange habanero chili peppers. J. Food Sci. 67, 956–962 (2002)CrossRefGoogle Scholar
  10. 10.
    Chanthai, S., Juangsamoot, J., Ruangviriyachai, C., Techawongstien, S.: Determination of capsaicin and dihydrocapsaicin in some chilli varieties using accelerated solvent extraction associated with solid-phase extraction methods and RP-HPLC-fluorescence. E- J. Chem. 9, 1550–1561 (2012)CrossRefGoogle Scholar
  11. 11.
    Reilly, C.A., Crouch, D.J., Yost, G.S., Fatah, A.A.: Determination of capsaicin, dihydrocapsaicin, and nonivamide in self-defense weapons by liquid chromatography-mass spectrometry and liquid chromatography-tandem mass spectrometry. J. Chromatogr. A 912, 259–267 (2001)CrossRefGoogle Scholar
  12. 12.
    Kozukue, N., Han, J.S., Kozukue, E., Lee, S.J., Kim, J.A., Lee, K.R., Levin, C.E., Friedman, M.: Analysis of eight capsaicinoids in peppers and pepper-containing foods by high-performance liquid chromatography and liquid chromatography-mass spectrometry. J. Agric. Food Chem. 53, 9172–9181 (2005)CrossRefGoogle Scholar
  13. 13.
    Laskaridou-Monnerville, A.: Determination of capsaicin and dihydrocapsaicin by micellar electrokinetic capillary chromatography and its application to various species of capscium, solanaceae. J. Chromatogr. A 838, 293–302 (1999)CrossRefGoogle Scholar
  14. 14.
    Liu, L., Chen, X., Liu, J., Deng, X., Duan, W., Tan, S.: Determination of capsaicin and dihydrocapsaicin in Capsicum anuum and related products by capillary electrophoresis with a mixed surfactant system. Food Chem. 119, 1228–1232 (2010)CrossRefGoogle Scholar
  15. 15.
    Kanu, A.B., Dwivedi, P., Tam, M., Matz, L., Hill Jr., H.H.: Ion mobility-mass spectrometry. J. Mass Spectrom. 43, 1–22 (2008)CrossRefGoogle Scholar
  16. 16.
    Wittmer, D., Luckenbill, B.K., Hill, H.H., Chen, Y.H.: Electrospray-ionization ion mobility spectrometry. Anal. Chem. 66, 2348–2355 (1994)CrossRefGoogle Scholar
  17. 17.
    Wu, C., Siems, W.F., Asbury, G.R., Hill, H.H.: Electrospray ionization high-resolution ion mobility spectrometry-mass spectrometry. Anal. Chem. 70, 4929–4938 (1998)CrossRefGoogle Scholar
  18. 18.
    Bohrer, B.C., Mererbloom, S.I., Koeniger, S.L., Hilderbrand, A.E., Clemmer, D.E.: Biomolecule analysis by ion mobility spectrometry. Annu. Rev. Anal. Chem. 1, 293–327 (2008)Google Scholar
  19. 19.
    Ewing, R.G., Atkinson, D.A., Eiceman, G.A., Ewing, G.J.: A critical review of ion mobility spectrometry for the detection of explosives and explosive related compounds. Talanta 54, 515–529 (2001)CrossRefGoogle Scholar
  20. 20.
    Weston, D.J., Bateman, R., Wilson, I.D., Wood, T.R., Creaser, C.S.: Direct analysis of pharmaceutical drug formulations using ion mobility spectrometry/quadrupole-time-of-flight mass spectrometry combined with desorption electrospray ionization. Anal. Chem. 77, 7572–7580 (2005)CrossRefGoogle Scholar
  21. 21.
    Dwivedi, P., Schultz, A.J., Hill Jr., H.H.: Metabolic profiling of human blood by high-resolution ion mobility mass spectrometry (IM-MS). Int. J. Mass Spectrom. 298, 78–90 (2010)CrossRefGoogle Scholar
  22. 22.
    Franceschi, P., Vrhovsek, U., Guella, G.: Ion mobility mass spectrometric investigation of ellagitannins and their noncovalent aggregates. Rapid Commun. Mass Spectrom. 25, 827–833 (2011)CrossRefGoogle Scholar
  23. 23.
    Gilar, M., Olivova, P., Daly, A.E., Gebler, J.C.: Orthogonality of separation in two-dimensional liquid chromatography. Anal. Chem. 77, 6426–6434 (2005)CrossRefGoogle Scholar
  24. 24.
    Imami, K., Sugiyama, N., Kyono, Y., Tomita, M., Ishirama, Y.: Automated phosphoproteome analysis for cultured cancer cells by two-dimensional nanoLC-MS using a calcined titania/C18 biphasic column. Anal. Sci. 24, 161–166 (2008)CrossRefGoogle Scholar
  25. 25.
    Peng, J.M., Elias, J.E., Thoreen, C.C., Licklider, L.J., Gygi, S.P.: Evaluation of multidimensional chromatography coupled with tandem mass spectrometry (LC/LC-MS/MS) for large-scale protein analysis: the yeast proteome. J. Proteome Res. 2, 43–50 (2003)CrossRefGoogle Scholar
  26. 26.
    Dugo, P., Cacciola, F., Kumm, T., Dugo, G., Mondello, L.: Comprehensive multidimensional liquid chromatography: theory and applications. J. Chromatogr. A 1184, 353–368 (2008)CrossRefGoogle Scholar
  27. 27.
    Gorg, A., Weiss, W., Dunn, M.J.: Current two-dimensional electrophoresis technology for proteomics. Proteomics 4, 3665–3685 (2004)CrossRefGoogle Scholar
  28. 28.
    Stoll, D.R., Li, X., Wang, X., Carr, P.W., Porter, S.E.G., Rutan, S.C.: Fast, comprehensive two-dimensional liquid chromatography. J. Chromatogr. A 1168, 3–43 (2007)CrossRefGoogle Scholar
  29. 29.
    Liu, X., Valentine, S.J., Plasencia, M.D., Trimpin, S., Naylor, S., Clemmer, D.E.: Mapping the human plasma proteome by SCX-LC-IMS-MS. J. Am. Soc. Mass Spectrom. 18, 1249–1264 (2007)CrossRefGoogle Scholar
  30. 30.
    Ma, Q., Wang, C., Bai, H., Xi, H.-W., Xi, G.-C., Ren, X.-M., Yang, Y., Guo, L.-H.: Comprehensive two-dimensional separation of hydroxylated polybrominated diphenyl ethers by ultra-performance liquid chromatography coupled with ion mobility-mass spectrometry. J. Am. Soc. Mass Spectrom. 22, 1851–1861 (2011)CrossRefGoogle Scholar
  31. 31.
    Ma, Q., Xi, G.-C., Wang, C., Bai, H., Zhang, Q., Xi, H.-W., Wang, Z.-M., Guo, L.-H.: Comprehensive two-dimensional separation for the analysis of alkylphenol ethoxylates employing hydrophilic interaction chromatography coupled with ion mobility-mass spectrometry. Int. J. Mass Spectrom. 315, 31–39 (2012)CrossRefGoogle Scholar
  32. 32.
    McLean, J.A., Ruotolo, B.T., Gillig, K.J., Russell, D.H.: Ion mobility-mass spectrometry: a new paradigm for proteomics. Int. J. Mass Spectrom. 240, 301–315 (2005)CrossRefGoogle Scholar
  33. 33.
    Varesio, E., Le Blanc, J.C.Y., Hopfgartner, G.: Real-time 2D separation by LC x differential ion mobility hyphenated to mass spectrometry. Anal. Bioanal. Chem. 402, 2555–2564 (2012)CrossRefGoogle Scholar
  34. 34.
    Li, H., Giles, K., Bendiak, B., Kaplan, K., Siems, W.F., Hill Jr., H.H.: Resolving structural isomers of monosaccharide methyl glycosides using drift tube and traveling wave ion mobility mass spectrometry. Anal. Chem. 84, 3231–3239 (2012)CrossRefGoogle Scholar
  35. 35.
    Kwasnik, M., Fernandez, F.M.: Theoretical and experimental study of the achievable separation power in resistive-glass atmospheric pressure ion mobility spectrometry. Rapid Commun. Mass Spectrom. 24, 1911–1918 (2010)CrossRefGoogle Scholar
  36. 36.
    Lapthorn, C., Pullen, F., Chowdhry, B.Z.: Ion mobility spectrometry-mass spectrometry (IMS-MS) of small molecules: separating and assigning structures to ions. Mass Spectrom. Rev. 32, 43–71 (2013)CrossRefGoogle Scholar
  37. 37.
    Yamaguchi, Y., Nishima, W., Re, S., Sugita, Y.: Confident identification of isomeric N-glycan structures by combined ion mobility mass spectrometry and hydrophilic interaction liquid chromatography. Rapid Commun. Mass Spectrom. 26, 2877–2884 (2012)CrossRefGoogle Scholar
  38. 38.
    Barnes, C.A.S., Hilderbrand, A.E., Valentine, S.J., Clemmer, D.E.: Resolving isomeric peptide mixtures: a combined HPLC/ion mobility-TOFMS analysis of a 4000-component combinatorial library. Anal. Chem. 74, 26–36 (2002)CrossRefGoogle Scholar
  39. 39.
    Tang, K., Shvartsburg, A.A., Lee, H.N., Prior, D.C., Buschbach, M.A., Li, F.M., Tolmachev, A.V., Anderson, G.A., Smith, R.D.: High-sensitivity ion mobility spectrometry/mass spectrometry using electrodynamic ion funnel interfaces. Anal. Chem. 77, 3330–3339 (2005)CrossRefGoogle Scholar
  40. 40.
    Belov, M.E., Clowers, B.H., Prior, D.C., Danielson III, W.F., Liyu, A.V., Petritis, B.O., Smith, R.D.: Dynamically multiplexed ion mobility time-of-flight mass spectrometry. Anal. Chem. 80, 5873–5883 (2008)CrossRefGoogle Scholar
  41. 41.
    Clowers, B.H., Belov, M.E., Prior, D.C., Danielson III, W.F., Ibrahim, Y., Smith, R.D.: Pseudorandom sequence modifications for ion mobility orthogonal time-of-flight mass spectrometry. Anal. Chem. 80, 2464–2473 (2008)CrossRefGoogle Scholar
  42. 42.
    Zhang, X., Siems, W.F., Graf, S., Knochenmuss, R., Hill, H.H. Jr.: Hadamard transform atmospheric pressure ion mobility time-of-flight mass spectrometry for complex sample analysis. Proceedings of the 61th ASMS Conference on Mass Spectrometry and Allied Topics, Minneapolis, MN, June 9−13 (2013) TP 625Google Scholar
  43. 43.
    Zhang, X., Chiu, V.M., Stoica, G., Gina, L., Schenk, J.O., Hill Jr., H.H.: Metabolic analysis of striatal tissues from parkinson's disease-like rats by electrospray ionization ion mobility mass spectrometry. Anal. Chem. 86, 3075–3083 (2014)CrossRefGoogle Scholar
  44. 44.
    Zhang, X., Knochenmuss, R., Siems, W.F., Liu, W., Graf, S., Hill Jr., H.H.: Evaluation of Hadamard transform atmospheric pressure ion mobility time-of-flight mass spectrometry for complex mixture analysis. Anal. Chem. 86, 1661–1670 (2014)CrossRefGoogle Scholar
  45. 45.
    Roscioli, K.M., Zhang, X., Li, S.X., Goetz, G.H., Cheng, G., Zhang, Z., Siems, W.F., Hill Jr., H.H.: Real time pharmaceutical reaction monitoring by electrospray ion mobility-mass spectrometry. Int. J. Mass Spectrom. 336, 27–36 (2013)CrossRefGoogle Scholar
  46. 46.
    Clowers, B.H., Siems, W.F., Hill, H.H., Massick, S.M.: Hadamard transform ion mobility spectrometry. Anal. Chem. 78, 44–51 (2006)CrossRefGoogle Scholar
  47. 47.
    Szumlas, A.W., Ray, S.J., Hieftje, G.M.: Hadamard transform ion mobility spectrometry. Anal. Chem. 78, 4474–4481 (2006)CrossRefGoogle Scholar
  48. 48.
    Knochenmuss, R., Graf, S., Fuhrer, K., Gonin, M. Proceedings of the 61th ASMS Conference on Mass Spectrometry and Allied Topics, Minneapolis, MN, June 9−13 (2013) WP 744Google Scholar
  49. 49.
    Xia, Y.-Q., Wu, S.T., Jemal, M.: LC-FAIMS-MS/MS for quantification of a peptide in plasma and evaluation of FAIMS global selectivity from plasma components. Anal. Chem. 80, 7137–7143 (2008)CrossRefGoogle Scholar
  50. 50.
    Klaassen, T., Szwandt, S., Kapron, J.T., Roemer, A.: Validated quantitation method for a peptide in rat serum using liquid chromatography/high-field asymmetric waveform ion mobility spectrometry. Rapid Commun. Mass Spectrom. 23, 2301–2306 (2009)CrossRefGoogle Scholar
  51. 51.
    Smith, R.W., Toutoungi, D.E., Reynolds, J.C., Bristow, A.W.T., Ray, A., Sage, A., Wilson, I.D., Weston, D.J., Boyle, B., Creaser, C.S.: Enhanced performance in the determination of ibuprofen 1-beta-O-acyl glucuronide in urine by combining high field asymmetric waveform ion mobility spectrometry with liquid chromatography-time-of-flight mass spectrometry. J. Chromatogr. A 1278, 76–81 (2013)CrossRefGoogle Scholar
  52. 52.
    Parastar, H., Radovic, J.R., Jalali-Heravi, M., Diez, S., Bayona, J.M., Tauler, R.: Resolution and quantification of complex mixtures of polycyclic aromatic hydrocarbons in heavy fuel oil sample by means of GC x GC-TOFMS combined to multivariate curve resolution. Anal. Chem. 83, 9289–9297 (2011)CrossRefGoogle Scholar
  53. 53.
    De la Mata, A.P., Harynuk, J.J.: Limits of detection and quantification in comprehensive multidimensional separations. 1. A theoretical look. Anal. Chem. 84, 6646–6653 (2012)CrossRefGoogle Scholar
  54. 54.
    Mondello, L., Herrero, M., Kumm, T., Dugo, P., Cortes, H., Dugo, G.: Quantification in comprehensive two-dimensional liquid chromatography. Anal. Chem. 80, 5418–5424 (2008)CrossRefGoogle Scholar
  55. 55.
    Valentine, S.J., Kulchania, M., Barnes, C.A.S., Clemmer, D.E.: Multidimensional separations of complex peptide mixtures: a combined high-performance liquid chromatography/ion mobility/time-of-flight mass spectrometry approach. Int. J. Mass Spectrom. 212, 97–109 (2001)CrossRefGoogle Scholar
  56. 56.
    Butnariu, M., Caunii, A., Putnoky, S.: Reverse phase chromatographic behaviour of major components in Capsicum annuum extract. Chem. Cent. J. 6, 146 (2012)CrossRefGoogle Scholar
  57. 57.
    Takahashi, M., Osawa, K., Tsuai, C.T., Abe, M.: Constituents of Capsicum-annuum-l var fasciculatum Irish and the distribution of capsi-amide in the varieties of Capsicum-annuum-l. Yakugaku Zasshi 100, 221–223 (1980)Google Scholar
  58. 58.
    Beegle, L.W., Kanik, I., Matz, L., Hill, H.H.: Electrospray ionization nigh-resolution ion mobility spectrometry for the detection of organic compounds, 1. Amino acids. Anal. Chem. 73, 3028–3034 (2001)CrossRefGoogle Scholar
  59. 59.
    Johnson, P.V., Kim, H.I., Beegle, L.W., Kanik, I.: Electrospray ionization ion mobility spectrometry of amino acids: ion mobilities and a mass-mobility correlation. J. Phys. Chem. A 108, 5785–5792 (2004)CrossRefGoogle Scholar

Copyright information

© American Society for Mass Spectrometry 2016

Authors and Affiliations

  • Wenjie Liu
    • 1
    • 2
  • Xing Zhang
    • 2
  • Richard Knochenmuss
    • 3
  • William F. Siems
    • 2
  • Herbert H. HillJr.
    • 2
    Email author
  1. 1.Key Laboratory of Protection and Utilization of Biological Resources in Tarim Basin of Xinjiang Production and Construction CorpsTarim UniversityAlarChina
  2. 2.Department of ChemistryWashington State UniversityPullmanUSA
  3. 3.Tofwerk, AGThunSwitzerland

Personalised recommendations