Investigating Differences in Gas-Phase Conformations of 25-Hydroxyvitamin D3 Sodiated Epimers using Ion Mobility-Mass Spectrometry and Theoretical Modeling

  • Christopher D. Chouinard
  • Vinícius Wilian D. Cruzeiro
  • Christopher R. Beekman
  • Adrian E. Roitberg
  • Richard A. Yost
Research Article


Drift tube ion mobility coupled with mass spectrometry was used to investigate the gas-phase structure of 25-hydroxyvitamin D3 (25OHD3) and D2 (25OHD2) epimers, and to evaluate its potential in rapid separation of these compounds. Experimental results revealed two distinct drift species for the 25OHD3 sodiated monomer, whereas only one of these conformations was observed for its epimer (epi25OHD3). The unique species allowed 25OHD3 to be readily distinguished, and the same pattern was observed for 25OHD2 epimers. Theoretical modeling of 25OHD3 epimers identified energetically stable gas-phase structures, indicating that both compounds may adopt a compact “closed” conformation, but that 25OHD3 may also adopt a slightly less energetically favorable “open” conformation that is not accessible to its epimer. Calculated theoretical collision cross-sections for these structures agreed with experimental results to <2%. Experimentation indicated that additional energy in the ESI source (i.e., increased temperature, spray voltage) affected the ratio of 25OHD3 conformations, with the less energetically favorable “open” conformation increasing in relative intensity. Finally, LC-IM-MS results yielded linear quantitation of 25OHD3, in the presence of the epimer interference, at biologically relevant concentrations. This study demonstrates that ion mobility can be used in tandem with theoretical modeling to determine structural differences that contribute to drift separation. These separation capabilities provide potential for rapid (<60 ms) identification of 25OHD3 and 25OHD2 in mixtures with their epimers.

Graphical Abstract


Ion mobility spectrometry Mass spectrometry Vitamin D Theoretical modeling Epimers 


The analysis of vitamin D metabolites has seen a remarkable increase in clinical laboratories over the last 25 years, due to their importance in bone health [1] and several other diseases, including cancer, diabetes, and cardiovascular conditions [2]. Vitamin D refers to a class of secosteroids, with parent compounds vitamin D3 and D2; D3 is photosynthesized in animals, whereas D2 is found naturally in plants and is often used as a supplement in food sources [1]. The parent compounds are converted to 25-hydroxyvitamin D3 (25OHD3) and D2 (25OHD2) (Figure 1) by the enzyme 25-hydroxylase. These compounds, although inactive, are the most abundant circulating vitamin D metabolites, and are thus the most common clinical target analytes.
Figure 1

Structures of 25-hydroxyvitamin D3 (left) and 25-hydroxyvitamin D2, indicating the major structural difference between D3 and D2 compounds (blue). Their respective epimers differ in the chirality of the C3 hydroxyl group (red)

Normal plasma concentrations range from approximately 20–60 ng/mL [2, 3, 4], which is up to several orders of magnitude higher than other vitamin D metabolites, such as the primary biologically active form 1,25-dihydroxyvitamin D. Traditionally, clinical analysis has been performed with immunoassays [5], which recently made up more than 90% of routine analyses [6]. However, liquid chromatography-tandem mass spectrometry (LC-MS/MS) methods have increased in usage due to the technique’s selectivity and sensitivity [7, 8].

LC-MS/MS analysis of 25OHD3 and 25OHD2 is not without its challenges; difficulties arise primarily from low ionization efficiency and interfering isobars and isomers [4, 9]. One isomeric interference of interest is the C3 epimer of each compound, 3-epi-25OHD3 (epi25OHD3) and 3-epi-25OHD2 (epi25OHD2). These compounds are enzymatically formed in a unidirectional (irreversible) process by 3(α,β)-hydroxysteroid epimerase [10]. Although these epimers are highest in concentration in infants [3], significant concentrations (ranging from 0.1 to 24 ng/mL) have also been identified in adults [11, 12]. As such, current methods that fail to differentiate 25OHD3 and 25OHD2 from their respective epimers will present positively biased results [13], not only by inclusion of the epimer in the quantitation but also due to the slightly higher ionization efficiency of the epimers [14]. Additionally, although the biochemical role of the C3 epimers is not fully understood, studies have indicated that the epimer compounds of the active 1,25-dihydroxyvitamin D compounds (which can be synthesized from epi25OHD or epimerized from 1,25-dihydroxyvitamin D) are less biologically active than their counterparts, with reduced calcemic and non-calcemic effects [3, 10].

Differentiation of the epimers with LC-MS/MS methods is not routine in clinical laboratories, as many clinicians choose not to distinguish these compounds [3, 13]. This is primarily because the epimers display similar MS/MS spectra and their separation requires extended chromatography times or specialty columns. Although numerous LC methods capable of chromatographic resolution have been published in the last 5 y [12, 15, 16, 17, 18, 19], time of analysis generally exceeds 10–20 min per sample. A more rapid method of separating these epimers would allow a reduced emphasis on chromatographic separation, ultimately decreasing overall time of analysis and improving throughput. One technique that offers promise in this regard is ion mobility spectrometry (IMS) [20, 21], which separates gas-phase ions based on differences in their size, shape, and charge. A distinct advantage to IMS is its short time of analysis; IMS acquisition cycles are typically on the order of <100 ms. When coupled with chromatography, IMS is therefore capable of providing several individual mobility spectra across a single chromatographic peak [22]. IMS can also be used as the primary separation tool with direct infusion studies [23]. When coupled with time-of-flight mass spectrometry, with acquisition time on the order of 100 μs, IMS allows accurate identification of each ion mobility peak by providing a complete mass spectrum [24]. This technique has been used historically for both defense applications (i.e., explosives and chemical warfare agent detection) [25, 26] and as a technique to study structure and folding of macromolecules (i.e., proteins) [27, 28, 29, 30]. It is also being increasingly used for small molecule biological applications, especially in separation of isomeric compounds that are difficult to resolve with LC-MS [31, 32, 33, 34].

Current commercial IMS instrumentation often lacks the performance necessary to routinely resolve small molecule isomers with minor structural differences, including stereoisomers. As such, numerous strategies have been employed to augment separation, including analysis of different ionization species (i.e., by addition of cations such as sodium, lithium, or potassium) [34, 35, 36, 37, 38, 39], use of alternative drift gas environments [40, 41], and supplementation with theoretical modeling to better elucidate experimental results and potentially provide predictive capabilities to improve the experimental process. Theoretical modeling has been implemented for decades to provide further insight into experimental results in gas-phase structural studies, and this process can be used to improve separation capabilities by increasing confidence in identification of conformers based on drift time differences [42, 43, 44]. This strategy has seen application across several groups of biomolecules, ranging from mononucleotides to peptides and proteins [27, 28, 39, 45, 46, 47], as well as to pharmaceutical drugs [48]. Our group has also recently employed modeling to explain significant structural differences observed in the ion mobility spectra of sodiated steroid dimers [49]. Theoretical modeling of three-dimensional structure allows prediction of an ion’s collision cross-section, or the effective area of a gas-phase ion that accounts for its size, shape, and charge [42, 43, 48, 50]. Cross-section can also be determined experimentally by measuring an ion’s drift time through a cell maintained at constant experimental conditions. Comparison of theoretically and experimentally derived cross-sections can often help to explain experimental results, and can also accurately determine gas-phase structures.

Utilization of IMS-based techniques has been demonstrated for several classes of small molecules, including steroid-based compounds [51, 52], but analysis of vitamin D metabolites by IMS has been limited [9]. Here we report a combined experimental and theoretical approach for the investigation of 25-hydroxyvitamin D metabolites, and specifically the separation capabilities afforded by the identification of unique gas-phase sodiated conformers.
$$ \varOmega =\frac{{\left(18\pi \right)}^{1/2}}{16}\frac{ze}{{\left({k}_B T\right)}^{1/2}}{\left[\frac{1}{m_I}+\frac{1}{m_B}\right]}^{1/2}\frac{t_d E}{L}\frac{760}{P}\frac{T}{273.2}\frac{1}{N} $$


Materials and Reagents

25-Hydroxyvitamin D3, 3-epi-25-hydroxyvitamin D3, 25-hydroxyvitamin D2, 3-epi-25-hydroxyvitamin D2, vitamin D3, vitamin D2, and D6-25-hydroxyvitamin D3 standards (100 μg/mL in ethanol) were purchased from IsoSciences (King of Prussia, PA, USA). These standards were diluted to a final concentration of 10 μg/mL in Fisher Optima LC-MS grade methanol, purchased from Fisher Scientific (Pittsburgh, PA, USA), with no additives.

IM-MS Analysis

All analyses were performed with an Agilent 6560 IM-QTOF (Santa Clara, CA, USA). Standard solutions were directly infused by syringe pump at a flow rate of 10 μL/min. Although the majority of clinical LC-MS/MS analyses use atmospheric pressure chemical ionization (APCI), our results indicated poor epimer separation for the ions formed with this method (Supplementary Figure S-1). As such, all compounds were analyzed in positive mode using an Agilent Jet Stream (AJS) electrospray ionization (ESI) source. The ESI source conditions were as follows: capillary voltage: +5000 V; nozzle voltage: +1000 V; drying gas: 325 °C at 5 L/min; sheath gas: 275 °C at 8 L/min. The IM- QTOF instrument consists of a 78 cm uniform field drift tube maintained at approximately 4 Torr nitrogen drift gas and 32 °C. These constant drift tube conditions allow direct comparison of drift time spectra, which were acquired over a 60 ms window.

A modified version of the Mason-Schamp equation (Equation 1) [20, 53, 54] was used to measure collision cross-section (CCS) for ions of interest, where ze is the charge, kB is the Boltzman constant, mI is the analyte ion mass, mB is the buffer gas molecule mass, td is the corrected drift time, E is the electric field strength in V · cm–1, L is the drift tube length in cm, P is the drift tube pressure in Torr, T is the drift tube temperature in Kelvin, and N is the drift tube number density in cm–3. To calculate the corrected drift time (td), the drift tube field was varied over eight field strength steps from 9.6 to 18.6 V/cm (drift tube entrance voltage 750–1450 V). Total drift time, tD, was plotted versus the inverse of the drift tube voltage and extrapolated to the y-intercept, which represents the non-drift tube time, t0. This value was then subtracted from the measured total drift time to yield the corrected drift time. All drift time spectra shown were acquired at 18.6 V/cm, unless otherwise noted; optimal peak resolving power was achieved at this field strength.

Time of flight mass spectra were acquired in full scan high resolution mode over a range from m/z 100 to 1700. All IM-MS data processing was performed using Agilent IM-MS Browser B.07.01. A schematic of the Agilent 6560 IM-QTOF instrument used is shown in Supplementary Figure S-2.

Theoretical Modeling Methods

Owing to the high degree of flexibility in 25-hydroxyvitamin D metabolites because of the presence of the carbon chain extending from the D-ring of the compounds (Figure 1), performing a conformational search in order to find the most stable conformation can be a challenging task. Hereafter we describe the procedure used to address this problem.

An initial gas-phase conformation for both 25OHD3 and epi25OHD3 sodiated monomers was optimized using the Gaussian 09 program [55] at the level B3LYP-D3/6-31G*, where D3 means that the Grimme’s empirical dispersion correction [56] was added to the B3LYP functional. These structures were then used to generate GAFF force field parameters [57] on AMBER 14 [58], where only Van der Waals and electrostatic interactions were considered on the sodium ion (therefore, the ion is not bonded to the molecule). These parameters were used to perform classic gas-phase molecular dynamics for 25OHD3 and epi25OHD3 sodiated monomers. The dynamics were performed on AMBER 14 [58] with a time step of 2 fs. After an initial minimization, the system was heated from 0 to 700 K during 0.2 ns. Afterwards, 1.8 ns of production were generated at constant temperature (700 K).

This molecular dynamics was only performed to generate different conformations for the 25OHD3 and epi25OHD3 sodiated monomers as initial structures for the conformational search. The high temperature (700 K) gives more freedom for the molecules to overcome conformational barriers, thus improving the sampling on the conformational search. One hundred snapshots equally spaced in time were taken from the production phase of both dynamics. Each snapshot was submitted to a B3LYP-D3/6-31G* in-water geometry optimization, where the SMD implicit solvation model designed by Truhlar and coworkers [59] was used to account for solvent effects. Vibrational frequency calculations were also performed at the same level in order to obtain the Gibbs free energy value at T = 300 K for each structure.

The most stable in-water structures were submitted to gas-phase geometry optimizations, and the resulting structures were submitted to gas-phase collision cross-section calculations using the trajectory method (TM) in the MOBCAL software package [43]. In these calculations, N2 was used as the drift gas. For each structure, the atomic point charges needed for CCS calculations with TM were obtained using CHelpG [60] charges fitted to the electrostatic potential calculated at the B3LYP-D3/6-31 + G(d,p) level using the Gaussian 09 program [55].

Results and Discussion

Comparison of Ion Mobility-Mass Spectrometry Results for the Epimers

Ion mobility-mass spectrometry data were collected individually for standard solutions (10 μg/mL in methanol, no additives) of 25-hydroxyvitamin D3 (25OHD3), 3-epi-25-hydroxyvitamin D3 (epi25OHD3), 25-hydroxyvitamin D2 (25OHD2), and 3-epi-25-hydroxyvitamin D2 (epi25OHD2). Structures for 25OHD3 and 25OHD2 are shown in Figure 1. 25OHD2 differs structurally from 25OHD3 only in the presence of a C22 double bond and a C24 methyl group, highlighted in blue. 25OHD3 and 25OHD2 both contain a C3 α-hydroxyl group (highlighted in red), whereas their respective epimers contain a C3 β-hydroxyl group. The full mass spectra showed that the major ion identified for all compounds was the sodiated monomer [M+Na]+ at m/z 423.324 (D3 compounds) and m/z 435.324 (D2 compounds); mass spectra for each individual compound are shown in Supplementary Figure S-3. Both 25OHD3 and 25OHD2 showed slightly lower overall intensity in the mass spectrum than their respective epimers, which agrees with previous literature [14, 61]. Additional ions were identified, including [2M+Na]+ and [3M+2Na]+2; however, the intensity of these species was considerably lower in the spectra for epi25OHD3 and epi25OHD2, and the [3M+2Na]+2 species was not detected for epi25OHD2. Owing to the low intensity, these additional species will not be considered further, but are briefly discussed in the Supplemental Information.

Ion mobility separation between the sodiated monomers for each epimer pair was evaluated, and overlays of the drift spectra for 25OHD3/epi25OHD3 and 25OHD2/epi25OHD2 are shown in Figure 2. Collision cross-sections were measured for all species, and are listed in Table 1. 25OHD3 exhibited major drift peaks at 25.99 and 29.47 ms (Figure 2a), but, in contrast, epi25OHD3 exhibited only one major drift peak at 25.78 ms. Although the intensity of both individual drift peaks for 25OHD3 was lower in intensity than the single peak for epi25OHD3, the sum of their intensities was only slightly less, which agrees with the experimental mass spectral intensities and previous literature indicating higher ionization efficiency for epi25OHD3. The mass spectrum for each of these drift species (Figure 2a inset) showed only a single peak at m/z 423.324, indicating that no other species besides the sodiated monomer were contributing to that mobility. Because of this observation, it was hypothesized that the two different peaks for 25OHD3 correspond to distinct gas-phase conformations of the monomer ion. This hypothesis will be discussed in greater detail in the next section, in which molecular modeling was performed to determine the most energetically favorable gas-phase conformation(s) for each of these compounds.
Figure 2

Overlay of sodiated monomer drift spectra for (a) 25OHD3 and epi25OHD3, and (b) 25OHD2 and epi25OHD2. Inset: mass spectra collected for (i) 25OHD3 at 26.0 ms, (ii) 25OHD3 at 29.0 ms, (iii) 25OHD2 at 26.0 ms, and (iv) 25OHD2 at 29.0 ms

Table 1

Measured Collision Cross Sections for each of the Identified Ionization Species of the Compounds. Values in (Parentheses) for 25OHD3 and 25OHD2 Sodiated Monomers Correspond to the Second Drift Peak

Ionization species

Experimental CCS (Å2)




[M+Na]+ m/z 423.324

207.1 ± 0.3 (232.9 ± 0.5)

205.1 ± 0.3




[M+Na]+ m/z 435.324

211.1 ± 0.5 (235.6 ± 0.5)

207.9 ± 0.3

The drift spectra obtained for the sodiated monomers of 25OHD2 epimers were also compared (Figure 2b). These compounds exhibited a similar pattern of drift peaks as the 25OHD3 epimers (two major peaks for 25OHD2, one for epi25OHD2). These peaks were similarly identified using mass spectrometry, and again the mass spectrum for each species (Figure 2b inset) showed only a single peak at m/z 435.324, indicating that no other species besides the sodiated monomer were contributing to that mobility.

Molecular Modeling Results

In order to explore the hypothesis of multiple 25OHD3 sodiated monomer conformers, a conformational search was performed (as described in the Theoretical Modeling Methods section) to identify the lowest energy conformation(s) for the 25OHD3 epimers (Figure 3). The “O1–Na bond – O2–Na bond” distance along the x-axis refers to the relative distance between the C3 and C25 hydroxyl group oxygen atoms in interaction with the sodium ion. A distance of zero yields a “closed” conformation where the carbon chain is bent such that both distal hydroxyl groups are in close interaction with the single sodium ion (Figure 4). We see that 25OHD3 and epi25OHD3 may adopt two conformations: “closed” (as explained previously), and “open”, where the sodium ion interacts exclusively with either one hydroxyl group or the other (“O1–Na bond – O2–Na bond” distance >> 0). Figure 4 shows the 3D structures of the most stable “closed” and “open” conformations of 25OHD3 and epi25OHD3.
Figure 3

Gibbs free energy at 300 K as a function of the difference between the two Na-O distances (a difference close to zero corresponds to a closed conformation). Points are shown for each structure obtained in the conformational search as described in the section “Theoretical Modeling Methods

Figure 4

Structures of the most stable “closed” and “open” conformations of 25OHD3 and epi25OHD3

For 25OHD3, the most stable structure found is the “closed” conformer. The most stable structure of the “open” conformer has the sodium ion interacting with the C3 hydroxyl group on the A-ring, and this structure is only 2.125 kcal/mol higher in Gibbs free energy than the most stable “closed” conformation. This minor free energy difference allows the “open” conformation to be formed, but in a lower abundance in comparison with the “closed” conformation. We see that the “open” conformations with the sodium ion placed near the C25 hydroxyl group on the chain are higher in Gibbs free energy, >6 kcal/mol higher than the most stable “closed” conformation, and therefore are much less likely to be formed.

For epi25OHD3, the most stable structure is also the “closed” conformer. The most stable “open” conformation (Na+ placed near the C3 hydroxyl on the A-ring) is 10.679 kcal/mol higher in Gibbs free energy than the most stable “closed” conformation. This large energy difference indicates that this conformer is not formed, and explains the experimental observation that the “open” conformation is not observed on the drift spectrum for epi25OHD3. None of the 100 structures selected from the molecular dynamics to be used in the conformational search were found to be “open” with the sodium interacting with the C25 hydroxyl group on the chain.

The behavior observed experimentally and theoretically for the “open” conformation can be attributed to the stereochemistry of the C3 hydroxyl group on the A-ring: the β-hydroxyl contained in epi25OHD3 allows sufficient bending of the molecule, such that the most energetically favorable ion assumes the “closed” conformation. However, the α-hydroxyl (bent on the other side of the A-ring) in 25OHD3 causes the minimum distance between the distal hydroxyl groups to be slightly larger, such that the attractive force is reduced, causing a significant percentage of these ions to remain in the “open” conformation.

Table 2 compares the experimentally measured and theoretically predicted collision cross-sections (CCS) for conformations of 25OHD3 and epi25OHD3. As discussed previously, for 25OHD3 the most stable “open” conformation has the sodium ion placed near the C3 hydroxyl group on the A-ring; we see in Table 2 that the CCS for this conformation agrees more closely with the experimental one than the CCS for the “open” conformation with the sodium near the chain. Overall, the theoretical CCS values agree to within <0.6% of the experimental ones, supporting that the theoretical approach used was able to successfully reproduce the experimental observations.
Table 2

Comparison of Experimental (CCSE) and Theoretical (CCST) Cross-Sections for 25OHD3 and epi25OHD3, with the Agreement Between Values in Percentage. Experimental Uncertainties and Theoretical Standard Deviations are also Shown




ΔCCS (%)




207.1 ± 0.3

206.2 ± 2.8



232.9 ± 0.5

231.7 ± 2.5



229.9 ± 2.3





205.1 ± 0.3

206.1 ± 1.8




221.3 ± 1.6



240.0 ± 3.5


Experimental Investigation of “Open” Conformer

To further investigate the unique 25OHD3 “open” conformer, a series of experiments was performed to investigate its properties: (1) evaluation of drift spectrum for epimer mixture; (2) examination of “open” conformer stability under varied experimental conditions; and (3) comparison of 25OHD3 cross-section with other vitamin D metabolites.
  1. (1)

    A mixture of deuterium-labeled 25OHD3 and epi25OHD3 was analyzed, and the nested IM-MS is shown in Supplementary Figure S-4. Note that deuterium-labeled 25OHD3 (D6-25OHD3) was used, to be differentiated by m/z from epi25OHD3. The figure demonstrates that in a mixture the sodiated monomer of 25OHD3 (deuterium-labeled at m/z 429.362) still displays the “open” conformation drift peak at approximately 29 ms, whereas the same peak is again absent for the sodiated monomer of epi25OHD3. This shows that 25OHD3 can be readily distinguished in a mixture, as would be encountered with biological samples, without interference from epi25OHD3.

  2. (2)

    The relative stability of the “open” conformer was monitored as a function of several instrumental parameters, by comparison of overall and relative IM intensity for the two drift peaks. Parameters included ionization source temperature (50–400 °C), capillary voltage (1500–5500 V), fragmentor voltage (200–600 V), octopole rf voltage (650–800 V), high pressure ion funnel rf voltage (100–200 V), trapping funnel rf voltage (100–200 V), drift tube field strength (9.6–18.6 V/cm), and rear funnel rf voltage (100–200 V); Supplementary Figure S-2 shows a block diagram of the instrument. The results indicate that the ratio of “open”:“closed” conformers was most affected during the process of converting ions from solution-phase to gas-phase. Specifically, ionization source parameters significantly changed the ratio of conformations: (a) as ionization source temperature was varied from 50 to 400 °C, the lower temperature (least energetic) source conditions (<250 °C) yielded an “open”:“closed” ratio <1, in agreement with energy calculations indicating the closed conformation is energetically more favorable; source temperatures above 300 °C yielded a ratio >1 (Supplementary Figure S-5). The increased temperature (increased energy) presumably allows a more energetically favorable conversion between conformations. Additional energy calculations were performed at an elevated temperature (650 K) similar to experimental conditions for each structure from the previous conformational search (Supplementary Figure S-6). Results showed that the energy difference was much closer to 0 kcal/mol at 650 K, in contrast to approximately 2 kcal/mol at 300 K. The theoretical trend of the “open” conformer becoming more favorable at higher temperature agrees with the experimental observation that the “open”:“closed” ratio increases. However, the same calculation for epi25OHD3 at 650 K revealed a Gibbs free energy difference of approximately 8 kcal/mol between “open” and “closed” conformers. This is in agreement with the experimental absence of the “open” conformation under all conditions tested. It should be noted that for 25OHD3 the free energy difference between the “closed” conformation and the lowest energy “open” conformation with Na+ on the C25 chain hydroxyl is only approximately 3 kcal/mol at 650 K (Supplementary Figure S-6), and as such one might expect to observe additional “open” conformations as unique drift species. However, the theoretical cross-sections for the lowest free energy structures of each “open” conformation are very similar (“Open”C3 = 231.7 Å2 versus “Open”C25 = 229.9 Å2), and so these conformers would not be expected to be fully resolved. The presence of multiple conformers may, however, contribute to the slightly larger peak width observed experimentally for the “open” conformation drift peak in comparison with the “closed” conformation peak (FWHM“Closed” = 0.50 ms versus FWHM“Open” = 0.63 ms). (b) Similarly, as ionization source voltage was varied from 1500 to 5500 V, the lowest voltage (least energetic) conditions resulted in an “open”:“closed” ratio <1, whereas higher voltages (≥5000 V) increased the relative ratio to >1 (Supplementary Figure S-5). The post-ionization source parameters had little effect on this ratio, indicating that conversion between conformations is unlikely once ions have transitioned from solution to gas phase. Modification of these source parameters to induce different conformations could also serve as a useful diagnostic tool.

  3. (3)

    25OHD3 and D2 were compared with their respective parent compounds, vitamin D3 and D2, which are identical in structure except for lack of C25 hydroxyl group. Presumably, lack of this hydroxyl group would yield an “open” conformation with the sodium interacting with the C3 hydroxyl group on the A-ring and no structural bend; therefore, this ion would be expected to have similar cross-section to the “open” conformation. Experimentally obtained CCS agreed to within 1% for each pair, with cross-sections of 231.1 ± 0.3 Å2 for vitamin D3 (ΔCCS25OHD3, “Open” = 0.8%) and 232.8 ± 0.3 Å2 for vitamin D2 (ΔCCS25OHD2, “Open” = 1.2%). Drift spectra for these compounds are shown in Supplementary Figure S-7.


Quantitative Analysis of 25OHD3 with LC-IM-MS

To determine the potential for IM-MS quantitation of 25OHD3 without interference from the epimer, an existing clinically validated LC-MS method [62] was modified to include IMS (i.e., LC-IM-MS). Other changes to the original method included use of ESI (rather than APCI) and QTOF mass spectrometry (rather than QQQ). The chromatographic conditions were reproduced and details can be found in the Supporting Information. Concentrations of 25OHD3 and epi25OHD3 ranging from 0 to 160 ng/mL were analyzed individually and in mixtures, and the intensity of the drift peak at 29 ms (corresponding to the “open” conformer of 25OHD3) was measured. The results yielded a linear calibration curve (Supplementary Figure S-8) for 25OHD3 over this concentration range. Furthermore, it was shown that quantitation of 25OHD3 was not affected by epi25OHD3, even at epimer concentrations as high as 80 ng/mL (Supplementary Figure S-9). The limit of detection (LOD) was determined to be 10 ng/mL, and because vitamin D deficiency and insufficiency are defined as <20 ng/mL and 21–29 ng/mL [1], respectively, this LC-IM-MS method is capable of quantitation below the deficiency threshold without interference from the epimer.

Conclusions and Broader Impact

This study, in an attempt to utilize ion mobility to separate 25-hydroxyvitamin D epimers, shows that the sodiated monomers of 25OHD3 and epi25OHD3 exhibit an overlapping drift peak at approximately 26 ms, but 25OHD3 also exhibits a unique peak at longer drift time (29 ms). This unique drift peak allows easy distinction of 25OHD3 in a mixture. Comparison of 25OHD2 and its epimer reveals the same experimental pattern. Theoretical modeling was used to determine that 25OHD3 may adopt two different energetically accessible conformations (“closed” and “open”), whereas for epi25OHD3 the “open” conformation is energetically not accessible. The theoretical modeling was able to explain all trends observed in the experimental drift spectra of 25OHD3 and epi25OHD3, and yielded theoretical CCS in agreement with the experimental values to within <0.6%. The occurrence of the “open” conformation creates a unique drift peak that can be utilized to conclusively identify 25OHD3 and 25OHD2, even in the presence of their respective epimers. Furthermore, this IMS method can be easily implemented with existing clinical LC-MS protocols and provide definitive quantitation of 25OHD3 below the deficiency threshold without interference from the epimer. This finding is expected to hold great potential for improving clinical analysis of 25-hydroxyvitamin D epimers, both by removing (or reducing) the required chromatographic resolution of the epimers and by effectively reducing the time of analysis, as IMS separation can be achieved in <60 ms.

Supporting Information

Results obtained with atmospheric pressure chemical ionization (APCI); instrument schematic; mass spectra of individual compounds; identification of post-drift tube fragmentation and other instrumental considerations; experimental investigation of “open” conformer; energy calculations performed at elevated temperature; drift spectra comparison for 25OHD metabolites and their respective parent compounds; and quantitative LC-IM-MS results for 25OHD3 at biologically relevant levels.



The authors gratefully acknowledge financial support from Agilent Technologies, Wellspring Clinical Lab, the University of Florida Graduate Fellowship, and CAPES (Brazil).

Supplementary material

13361_2017_1673_MOESM1_ESM.docx (511 kb)
ESM 1 Results obtained with atmospheric pressure chemical ionization (APCI); instrument schematic; mass spectra of individual compounds; identification of post-drift tube fragmentation and other instrumental considerations; experimental investigation of “open” conformer; energy calculations performed at elevated temperature; drift spectra comparison for 25OHD metabolites and their respective parent compounds; and quantitative LC-IM-MS results for 25OHD3 at biologically relevant levels. (DOCX 510 kb)


  1. 1.
    Holick, M.F.: Vitamin D deficiency. N. Engl. J. Med. 357, 266–281 (2007)CrossRefGoogle Scholar
  2. 2.
    Holick, M.F.: Vitamin D status: measurement, interpretation, and clinical application. Ann. Epidemiol. 19, 73–78 (2009)CrossRefGoogle Scholar
  3. 3.
    Bailey, D., Veljkovic, K., Yazdanpanah, M., Adeli, K.: Analytical measurement and clinical relevance of vitamin D3 C3-epimer. Clin. Biochem. 46, 190–196 (2013)CrossRefGoogle Scholar
  4. 4.
    Volmer, D.A., Mendes, L.R.B.C., Stokes, C.S.: Analysis of vitamin D metabolic markers by mass spectrometry: current techniques, limitations of the “gold standard” method, and anticipated future directions. Mass Spectrom. Rev. 34, 2–23 (2015)CrossRefGoogle Scholar
  5. 5.
    Heijboer, A.C., Blankenstein, M.A., Kema, I.P., Buijs, M.M.: Accuracy of six routine 25-hydroxyvitamin D assays: influence of vitamin D binding protein concentration. Clin. Chem. 58, 543–548 (2012)CrossRefGoogle Scholar
  6. 6.
    Muller, M.J., Volmer, D.A.: Mass spectrometric profiling of vitamin D metabolites beyond 25-hydroxyvitamin D. Clin. Chem. 61, 1033–1048 (2015)CrossRefGoogle Scholar
  7. 7.
    Shepard, R.M., Horst, R.L., Hamstra, A.J., Deluca, H.F.: Determination of vitamin D and its metabolites in plasma from normal and anephric man. Biochem. J. 182, 55–69 (1979)CrossRefGoogle Scholar
  8. 8.
    El-Khoury, J.M., Reineks, E.Z., Wang, S.: Progress of liquid chromatography-mass spectrometry in measurement of vitamin D metabolites and analogues. Clin. Biochem. 44, 66–76 (2011)CrossRefGoogle Scholar
  9. 9.
    Qi, Y., Geib, T., Schorr, P., Meier, F., Volmer, D.A.: On the isobaric space of 25-hydroxyvitamin D in human serum: potential for interferences in liquid chromatography/tandem mass spectrometry, systematic errors and accuracy issues. Rapid Commun. Mass Spectrom. 29, 1–9 (2014)CrossRefGoogle Scholar
  10. 10.
    Kamao, M., Tatematsu, S., Hatakeyama, S., Sakaki, T., Sawada, N., Inouye, K., Ozono, K., Kubodera, N., Reddy, G.S., Okano, T.: C-3 epimerization of vitamin D 3 metabolites and further metabolism of C-3 epimers. J. Biol. Chem. 279, 15897–15907 (2004)CrossRefGoogle Scholar
  11. 11.
    Stepman, H.C.M., Vanderroost, A., Stockl, D., Thienpont, L.M.: Full-scan mass spectral evidence for 3-epi-25-hydroxyvitamin D3 in serum of infants and adults. Clin. Chem. Lab. Med. 49, 253–256 (2011)Google Scholar
  12. 12.
    Lensmeyer, G., Poquette, M., Wiebe, D., Binkley, N.: The C-3 epimer of 25-hydroxyvitamin D 3 is present in adult serum. J. Clin. Endocrinol. Metab. 97, 163–168 (2012)CrossRefGoogle Scholar
  13. 13.
    Singh, R.J., Taylor, R.L., Reddy, G.S., Grebe, S.K.G.: C-3 epimers can account for a significant proportion of total circulating 25-hydroxyvitamin D in infants, complicating accurate measurement and interpretation of vitamin D status. J. Clin. Endocrinol. Metab. 91, 3055–3061 (2006)CrossRefGoogle Scholar
  14. 14.
    Van den Ouweland, J.M.W., Beijers, A.M., Van Daal, H.: Overestimation of 25-hydroxyvitamin D3 by increased ionisation efficiency of 3-epi-25-hydroxyvitamin D3 in LC-MS/MS methods not separating both metabolites as determined by an LC-MS/MS method for separate quantification of 25-hydroxyvitamin D3, 3-epi-25-hydroxyvitamin D3 and 25-hydroxyvitamin D2 in human serum. J. Chromatogr. B 967, 195–202 (2014)CrossRefGoogle Scholar
  15. 15.
    Shah, I., James, R., Barker, J., Petroczi, A., Naughton, D.P.: Misleading measures in Vitamin D analysis: a novel LC-MS/MS assay to account for epimers and isobars. Nutr. J. 10, 46 (2011)CrossRefGoogle Scholar
  16. 16.
    Bedner, M., Phinney, K.W.: Development and comparison of three liquid chromatography-atmospheric pressure chemical ionization/mass spectrometry methods for determining vitamin D metabolites in human serum. J. Chromatogr. A 1240, 132–139 (2012)CrossRefGoogle Scholar
  17. 17.
    Higashi, T., Suzuki, M., Hanai, J., Inagaki, S., Min, J.Z., Shimada, K., Toyo’oka, T.: A specific LC/ESI-MS/MS method for determination of 25-hydroxyvitamin D3 in neonatal dried blood spots containing a potential interfering metabolite, 3-epi-25-hydroxyvitamin D3. J. Sep. Sci. 34, 725–732 (2011)CrossRefGoogle Scholar
  18. 18.
    Van Den Ouweland, J.M.W., Beijers, A.M., Daal, H.V.: Fast separation of 25-hydroxyvitamin D3 from 3-epi-25-hydroxyvitamin D3 in human serum by liquid chromatography-tandem mass spectrometry: variable prevalence of 3-epi-25- hydroxyvitamin D3 in infants, children, and adults. Clin. Chem. 57, 1618–1619 (2011)CrossRefGoogle Scholar
  19. 19.
    Schleicher, R.L., Encisco, S.E., Chaudhary-Webb, M., Paliakov, E., McCoy, L.F., Pfeiffer, C.M.: Isotope dilution ultra performance liquid chromatography-tandem mass spectrometry method for simultaneous measurement of 25-hydroxyvitamin D2, 25-hydroxyvitamin D3 and 3-epi-25-hydroxyvitamin D3 in human serum. Clin. Chim. Acta 412, 1594–1599 (2011)CrossRefGoogle Scholar
  20. 20.
    Mason, E.A., Schamp, H.W.: Mobility of gaseous ions in weak electric fields. Ann. Phys. 4, 233–270 (1958)CrossRefGoogle Scholar
  21. 21.
    Cohen, M.J., Karasek, F.W.: Plasma Chromatography – a new dimension for gas chromatography and mass spectrometry. J. Chromatogr. Sci. 8, 330–337 (1970)CrossRefGoogle Scholar
  22. 22.
    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
  23. 23.
    Dwivedi, P., Wu, P., Klopsch, S.J., Puzon, G.J., Xun, L., Hill, H.H.: Metabolic profiling by ion mobility mass spectrometry (IMMS). Metabolomics 4, 63–80 (2008)CrossRefGoogle Scholar
  24. 24.
    Hoaglund, C.S., Valentine, S.J., Sporleder, C.R., Reilly, J.P., Clemmer, D.E.: Three-dimensional ion mobility/TOFMS analysis of electrosprayed biomolecules. Anal. Chem. 70, 2236–2242 (1998)CrossRefGoogle Scholar
  25. 25.
    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
  26. 26.
    Eiceman, G.A., Karpas, Z.: Ion mobility spectrometry. Taylor and Francis, New York (2005)CrossRefGoogle Scholar
  27. 27.
    Von Helden, G., Wyttenbach, T., Bowers, M.T.: Conformation of macromolecules in the gas phase: use of matrix-assisted laser desorption methods in ion chromatography. Science 267, 1483–1485 (1995)CrossRefGoogle Scholar
  28. 28.
    Clemmer, D.E., Hudgins, R.R., Jarrold, M.F.: Naked protein conformations: cytochrome c in the gas phase. J. Am. Chem. Soc. 117, 10141–10142 (1995)CrossRefGoogle Scholar
  29. 29.
    Valentine, S.J., Counterman, A.E., Clemmer, D.E.: Conformer-dependent proton-transfer reactions of ubiquitin ions. J. Am. Soc. Mass Spectrom. 8, 954–961 (1997)CrossRefGoogle Scholar
  30. 30.
    Shelimov, K.B., Jarrold, M.F.: Conformations, unfolding, and refolding of apomyoglobin in vacuum: an activation barrier for gas-phase protein folding. J. Am. Chem. Soc. 119, 2987–2994 (1997)CrossRefGoogle Scholar
  31. 31.
    Fenn, L.S., Kliman, M., Mahsut, A., Zhao, S.R., McLean, J.A.: Characterizing ion mobility-mass spectrometry conformation space for the analysis of complex biological samples. Anal. Bioanal. Chem. 394, 235–244 (2009)CrossRefGoogle Scholar
  32. 32.
    Fenn, L.S., McLean, J.A.: Structural resolution of carbohydrate positional and structural isomers based on gas-phase ion mobility-mass spectrometry. Phys. Chem. Chem. Phys. 13, 2196–2205 (2011)CrossRefGoogle Scholar
  33. 33.
    Srebalus Barnes, C.A., 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
  34. 34.
    Paglia, G., Kliman, M., Claude, E., Geromanos, S., Astarita, G.: Applications of ion-mobility mass spectrometry for lipid analysis. Anal. Bioanal. Chem. 407, 4995–5007 (2015)CrossRefGoogle Scholar
  35. 35.
    Carlesso, V., Afonso, C., Fournier, F., Tabet, J.C.: Stereochemical effects from doubly-charged iron clusters for the structural elucidation of diastereomeric monosaccharides using ESI/IT-MS. Int. J. Mass Spectrom. 219, 559–575 (2002)CrossRefGoogle Scholar
  36. 36.
    Clowers, B.H., Hill, H.H.: Influence of cation adduction on the separation characteristics of flavonoid diglycoside isomers using dual gate-ion mobility-quadrupole ion trap mass spectrometry. J. Mass Spectrom. 41, 339–351 (2006)CrossRefGoogle Scholar
  37. 37.
    Domalain, V., Tognetti, V., Hubert-Roux, M., Lange, C.M., Joubert, L., Baudoux, J., Rouden, J., Afonso, C.: Role of cationization and multimers formation for diastereomers differentiation by ion mobility-mass spectrometry. J. Am. Soc. Mass Spectrom. 24, 1437–1445 (2013)CrossRefGoogle Scholar
  38. 38.
    Jackson, P., Attalla, M.I.: N-nitrosopiperazines form at high pH in post-combustion capture solutions containing piperazine: a low-energy collisional behaviour study. Rapid Commun. Mass Spectrom. 24, 3567–3577 (2010)CrossRefGoogle Scholar
  39. 39.
    Leavell, M.D., Gaucher, S.P., Leary, J.A., Taraszka, J.A., Clemmer, D.E.: Conformational studies of Zn-ligand-hexose diastereomers using ion mobility measurements and density functional theory calculations. J. Am. Soc. Mass Spectrom. 13, 284–293 (2002)CrossRefGoogle Scholar
  40. 40.
    Dwivedi, P., Wu, C., Matz, L.M., Clowers, B.H., Siems, W.F., Hill Jr., H.H.: Gas-phase chiral separations by ion mobility spectrometry. Anal. Chem. 78, 8200–8206 (2006)CrossRefGoogle Scholar
  41. 41.
    Asbury, G., Hill, H.: Using different drift gases to change separation factors (alpha) in ion mobility spectrometry. Anal. Chem. 72, 580–4 (2000)CrossRefGoogle Scholar
  42. 42.
    Kim, H., Kim, H.I., Johnson, P.V., Beegle, L.W., Beauchamp, J.L., Goddard, W.A., Kanik, I.: Experimental and theoretical investigation into the correlation between mass and ion mobility for choline and other ammonium cations in N2. Anal. Chem. 80, 1928–1936 (2008)CrossRefGoogle Scholar
  43. 43.
    Mesleh, M.F., Hunter, J.M., Shvartsburg, A.A., Schatz, G.C., Jarrold, M.F.: Structural information from ion mobility measurements: effects of the long-range potential. J. Phys. Chem. A 101, 968–968 (1997)CrossRefGoogle Scholar
  44. 44.
    Zakharova, N.L., Crawford, C.L., Hauck, B.C., Quinton, J.K., Seims, W.F., Hill, H.H., Clark, A.E.: An assessment of computational methods for obtaining structural information of moderately flexible biomolecules from ion mobility spectrometry. J. Am. Soc. Mass Spectrom. 23, 792–805 (2012)CrossRefGoogle Scholar
  45. 45.
    Wyttenbach, T., Witt, M., Bowers, M.T.: On the stability of amino acid zwitterions in the gas phase: the influence of derivatization, proton affinity, and alkali ion addition. J. Am. Chem. Soc. 122, 3458–3464 (2000)CrossRefGoogle Scholar
  46. 46.
    Wyttenbach, T., Witt, M., Bowers, M.T.: On the question of salt bridges of cationized amino acids in the gas phase: glycine and arginine. Int. J. Mass Spectrom. 182/183, 243–252 (1999)CrossRefGoogle Scholar
  47. 47.
    Flick, T.G., Campuzano, I.D.G., Bartberger, M.D.: Structural resolution of proline diastereomers with ion mobilityspectrometry via alkali metal ion cationization. Anal. Chem. 87, 3300–3307 (2015)CrossRefGoogle Scholar
  48. 48.
    Campuzano, I., Bush, M.F., Robinson, C.V., Beaumont, C., Richardson, K., Kim, H., Kim, H.I.: Structural characterization of drug-like compounds by ion mobility mass spectrometry: comparison of theoretical and experimentally derived nitrogen collision cross sections. Anal. Chem. 84, 1026–1033 (2012)CrossRefGoogle Scholar
  49. 49.
    Chouinard, C.D., Cruzeiro, V.W.D., Roitberg, A., Yost, R.A.: Experimental and theoretical investigation of sodiated multimers of steroid epimers with ion mobility-mass spectrometry. J. Am. Soc. Mass Spectrom. 23, 323–331 (2017)CrossRefGoogle Scholar
  50. 50.
    Shvartsburg, A.A., Schatz, G.C., Jarrold, M.F.: Mobilities of carbon cluster ions: critical importance of the molecular attractive potential. J. Chem. Phys. 108, 2416 (1998)CrossRefGoogle Scholar
  51. 51.
    Chouinard, C.D., Beekman, C.R., Kemperman, R.H.J., King, H.M., Yost, R.A.: Ion mobility-mass spectrometry separation of steroid structural isomers and epimers. Int. J. Ion Mobil. Spectrom. (2016). doi: 10.1007/s12127-016-0213-4 Google Scholar
  52. 52.
    Ahonen, L., Fasciotti, M., Gennas, G.B., Kotiaho, T., Daroda, R.J., Eberlin, M., Kostiainen, R.: Separation of steroid isomers by ion mobility mass spectrometry. J. Chromatogr. A 1310, 133–137 (2013)CrossRefGoogle Scholar
  53. 53.
    Revercomb, H.E., Mason, E.A.: Theory of plasma chromatography/gaseous electrophoresis- a review. Anal. Chem. 47, 970–983 (1975)CrossRefGoogle Scholar
  54. 54.
    Mason, E.A., McDaniel, E.W.: Transport properties of ions in gases. Wiley, New York (1988)CrossRefGoogle Scholar
  55. 55.
    Frisch, M.J., Trucks, G.W., Schlegel, H.B., Scuseria, G.E., Robb, M.A., Cheeseman, J.R., Scalmani, G., Barone, V., Mennucci, B., Petersson, G.A., Nakatsuji, H., Caricato, M., Li, X., Hratchian, H.P., Izmaylov, A.F., Bloino, J., Zheng, G., Sonnenberg, J.L., Hada, M., Ehara, M., Toyota, K., Fukuda, R., Hasegawa, J., Ishida, M., Nakajima, T., Honda, Y., Kitao, O., Nakai, H., Vreven, T., Montgomery, J.A., Jr., Peralta, J.E., Ogliaro, F., Bearpark, M., Heyd, J.J., Brothers, E., Kudin, K.N., Staroverov, V.N., Kobayashi, R., Normand, J., Raghavachari, K., Rendell, A., Burant, J.C., Iyengar, S.S., Tomasi, J., Cossi, M., Rega, N., Millam, J.M., Klene, M., Knox, J.E., Cross, J.B., Bakken, V., Adamo, C., Jaramillo, J., Gomperts, R., Stratmann, R.E., Yazyev, O., Austin, A.J., Cammi, R., Pomelli, C., Ochterski, J.W., Martin, R.L., Morokuma, K., Zakrzewski, V.G., Voth, G.A., Salvador, P., Dannenberg, J.J., Dapprich, S., Daniels, A.D., Farkas, O., Foresman, J.B., Ortiz, J. V, Cioslowski, J., Fox, D.J.: Gaussian 09 Revision A.02 (2009)Google Scholar
  56. 56.
    Grimme, S., Antony, J., Ehrlich, S., Krieg, H.: A consistent and accurate ab initio parametrization of density functional dispersion correction (DFT-D) for the 94 elements H-Pu. J. Chem. Phys. 132, 154104 (2010)CrossRefGoogle Scholar
  57. 57.
    Wang, J., Wolf, R.M., Caldwell, J.W., Kollman, P.A., Case, D.A.: Development and testing of a general amber force field. J. Comput. Chem. 25, 1157–74 (2004)CrossRefGoogle Scholar
  58. 58.
    Case, D.A., Babin, V., Berryman, J.T., Betz, R.M., Cai, Q., Cerutti, D.S., Cheatham, T.E., III, Darden, T.A., Duke, R.E., Gohlke, H., Goetz, A.W., Gusarov, S., Homeyer, N., Janowski, P., Kaus, J., Kolossváry, I., Kovalenko, A., Lee, T.S., LeGrand, S., Luchko, T., Luo, R., Madej, B., Merz, K.M., Paesani, F., Roe, D.R., Roitberg, A., Sagui, C., Salomon-Ferrer, R., Seabra, G., Simmerling, C.L., Smith, W., Swails, J., Walker, R.C., Wang, J., Wolf, R.M., Wu, X., Kollman, P.A.: AMBER 14 (2014)Google Scholar
  59. 59.
    Marenich, A.V., Cramer, C.J., Truhlar, D.G.: Universal solvation model based on solute electron density and on a continuum model of the solvent defined by the bulk dielectric constant and atomic surface tensions. J. Phys. Chem. B 113, 6378–6396 (2009)CrossRefGoogle Scholar
  60. 60.
    Breneman, C.M., Wiberg, K.B.: Determining atom-centered monopoles from molecular electrostatic potentials. The need for high sampling density in formamide conformational analysis. J. Comput. Chem. 11, 361–373 (1990)CrossRefGoogle Scholar
  61. 61.
    Flynn, N., Lam, F., Dawnay, A.: Enhanced 3-epi-25-hydroxyvitamin D 3 signal leads to overestimation of its concentration and amplifies interference in 25-hydroxyvitamin D LC-MS/MS assays. Ann. Clin. Biochem. 51, 1–8 (2013)Google Scholar
  62. 62.
    Lee, D., Garrett, T.J., Goldberger, B.A., Bazydlo, L.A.L.: Quantitation of 25-hydroxyvitamin D2 and D3 in serum and plasma by LCMS/MS. Bioanalysis 7, 167–178 (2015)CrossRefGoogle Scholar

Copyright information

© American Society for Mass Spectrometry 2017

Authors and Affiliations

  • Christopher D. Chouinard
    • 1
  • Vinícius Wilian D. Cruzeiro
    • 1
    • 2
  • Christopher R. Beekman
    • 1
  • Adrian E. Roitberg
    • 1
  • Richard A. Yost
    • 1
    • 3
  1. 1.Department of ChemistryUniversity of FloridaGainesvilleUSA
  2. 2.CAPES FoundationMinistry of Education of BrazilBrasíliaBrazil
  3. 3.Southeast Center for Integrated Metabolomics (SECIM)University of FloridaGainesvilleUSA

Personalised recommendations