BIOMEDICAL CHROMATOGRAPHY, VOL. 5,153-160 (1991)
Analysis of Vitamin D and its Metabolites using Thermospray Liquid Chromatography/Mass Spectrometry David Watson and Kenneth D. R. Setchell* Mass Spectrometry Laboratory, Children’s Hospital Medical Center, Elland and Bethesda Avenues. Cincinnati, Ohio 45229, USA
Richardus Ross Department of Neonatology, Children’s Hospital Medical Center, Elland and Bethesda Avenues, Cincinnati, Ohio 45229, USA
A new method is described for the analysis of vitamin D and its metabolites utilizing thermospray (TSP) mass spectrometry as an on-line detector for high performance liquid chromatography. Ionization conditions were optimized for use with isocratic reversed phase chromatography. TSP mass spectrometry was employed in series with a W absorbance detector to facilitate comparisons between the two methods of detection. Positive ion TSP mass spectra were recorded for vitamin D2, vitamin D3, 25-hydroxyvitamin D3 (25(0H)D3), 1,25dihydroxyvitamin D, (1,25(OH),D3) and 24,25-dihydroxyvitamin D3 (24,25(OH),D,). The spectra contained protonated molecular ions, ammonium adduct ions and fragment ions due to the loss of one or more molecules of water. A comparison of quantitative precision was made by determining UV absorbance and TSP standard curves for vitamin D, using two different methods: (1)External standard method with post-column (post UV detector) addition of ammonium acetate. (2) As (1) but using the method of internal standards with a closely eluting internal standard (vitamin Dz). In each case the quantitative precision (correlation coefficient) for UV absorbance detection was superior owing to intrinsic instability of the TSP ion beam. A stable isotopically labelled internal standard was employed in the development of an assay for 1,25(OH),D,. The assay was used to quantify in uitro enzymic conversion of 25(OH)D3 to 1,25(oH),D3 in guinea pig and sheep renal mitochondria1 incubations. TSP LC/MS was also applied to analysis of an extract of human blood plasma in which D3and each of its principal metabolites were identified in a single analysis.
INTRODUCTION The metabolism of vitamin D and the biological activities of vitamin D metabolites have been studied with growing interest during the past 25 years. In order to facilitate clinical and biochemical investigations much effort has focussed on the development of appropriate analytical methods. Currently the methods routinely used for quantification of vitamin D3 metabolites include competitive protein binding assays (Belsey et al., 1971; Horst et al., 198l), radioreceptor assays (Chandler et al., 1980; Oftebro et af., 1988), radioimmonoassays (Gray et al., 1981; Hollis and Napoli, 1985) and high performance liquid chromatography (HPLC) with ultraviolet (UV) absorbance detection (Clemens et al., 1982; Okano et al., 1981). HPLC has also been employed in conjunction with competitive binding assays (Asknes, 1980; Jongen et al., 1981) to furnish greater compound specificity and to facilitate multicomponent analysis. As a result of the commercialization of the thermospray (TSP) interface and ion source (Blakley and Vestal, 1983) it is now routinely possible to introduce the entire eluate from analytical scale HPLC systems into a mass spectrometer without loss of chromato* Author to whom correspondence should be addressed. 0269-3879/91/040153-08 $05.00 01991 by John Wiley & Sons, Ltd.
graphic resolution, and on-line mass spectrometric analysis of polar and thermally labile compounds eluting from the HPLC column can be readily accomplished. Biomedical applications of TSP liquid chromatography/mass spectrometry (LC/MS) are now numerous, including analyses of drugs and drug metabolites (Covey et af., 1985), steroid conjugates (Watson et nl., 1985, 1986) corticosteroid hormones (Watson et al., 1987; Gaskell et al., 1987), eicosanoids (Richmond et al., 1986; Voyksner and Bush, 1987), phospholipids (Kim et al., 1987), glutathione conjugates (Parker et al., 1988), phytoestrogens (Setchell et al., 1987), bacterial phenazines (Watson et af., 1988), bile acids (Setchell and Vestal, 1989) and many others. Since HPLC is by far the most popular chromatographic technique in use for the purification and assay of vitamin D and its metabolites there is much potential benefit to be gained from the application of TSP mass spectrometry as an HPLC detector for these compounds. In concept TSP LC/MS possesses characteristics which should afford several advantages for analysis of vitamin D and its metabolites compared with gas chromatography/mass spectrometry (GUMS); it should be possible to reduce the number of sample extraction steps, no derivatization is required prior to chromatography and thermal effects should be eliminated. For these reasons we have investigated the qualitative and quantitative aspects of the TSP mass spectrometric behaviour of the most biologically important vitamin D related compounds. Received 21 August I990 Accepted I1 September I990
154
D. WATSON, R. ROSS AND K. D. R. SETCHELL
EXPERIMENTAL Chemicals. Vitamins D2 and D3 were purchased from Sigma Chemical Co. (St Louis, MO, USA). Vitamin D metabolites were generously donated by Hoffmann-la-Roche, Nutley, NJ, USA. These materials were stored in ethanol at -20 “C when not in use and the concentrations of stock solutions were verified by UV absorbance prior to use in quantitative analyses. Water for HPLC was purified to 18 M B cm resistivity; HPLC quality methanol was purchased from Baxter Healthcare Corporation, Muskegon, MI, USA. Ammonium acetate (HPLC grade) was obtained from J. T. Baker Chemical Co., Phillipsburg, NJ, USA. All other chemicals and biochemicals were standard commercial high purity preparations. Instrumentation. The HPLC mobile phase was delivered by a Waters Model 600-MS pump and the eluate was passed through the high pressure flow-cell of a Waters Model 490MS variable wavelength detector (set to monitor at 264 nm) connected directly to the vaporizer tube of a Vestec 201 TSP LClMS system (Vestec Corp., Houston, TX, USA). This system has a single quadrupole mass analyser (10-800 Da, unit resolution). All of the analyses were performed using a standard Vestec vaporizer tube with a 150 pm spray orifice. Samples were introduced into the mass spectrometer either via an HPLC column or using a column bypass loop with a Waters Model U6K injector. Instrument control and data acquisition were performed using a Teknivent Vector 1 (Teknivent, St Louis, MO, USA) mass spectrometer data system. High performance liquid chromatography. Three different reversed phase chromatography systems were developed in the course of these studies; while each was designed partly to satisfy the operating requirements of the TSP ionization process, the elution conditions are compatible with those used generally in the field of vitamin D research: Method A. For separation of hydroxylated vitamin D metabolites a Waters Nova-PAK c 1 8 column (15 cm X 0.46 cm i.d., 4 pm particle size) was eluted isocratically with methano1:water (80:20 v/v) containing ammonium acetate (0.1 M) at a flow rate of 1 mL/min. Method B. For Dz and D3 analysis a Hypersil CIRcolumn (25 cm X 0.46 cm i.d., 5 pm particle size, Keystone Scientific Inc., Bellefonte, PA, USA) was eluted with methanol (0.9 mL/min). The ionizing buffer for the TSP source (0.4 M ammonium acetate, pH 6.5, flow rate 0.3 mL/min) was added via a Valco zero dead volume “tee” connector on the outlet side of the UV absorbance detector using a separate pulsedampled HPLC pump (Waters Model 590). Method C. For simultaneous analysis of vitamins D2, D3 and their hydroxylated metabolites a combination of Methods A and B were used; the Hypersil column was eluted at 0.9 mllmin with a linear solvent gradient running from methanokwater (80:20 vlv) to 100% methanol over a 15-min period. The final solvent composition was held for 10 min at the end of the gradient. The ionizing buffer was added as in Method B. TSP mass spectra. The operating parameters of the TSP LC/MS system (vaporizer control temperature, vaporizer tip temperature, ion source block temperature) were initially optimized using background ions generated by ionization of HPLC Method A mobile phase which was continuously pumped into the ion source at a flow rate of lmL/min.
Ionization was enhanced when the ion source filament was switched on. Detailed optimization of ionization was then carried out using a “tuning solution” (a 2 pg/mL solution of 1,25(OH)2D3in the mobile phase, pumped continuously via the column bypass loop into the ion source at a flowrate of lmL/min). All available modes of ionization (pure TSP, “filament-ion’’ and “discharge-on”) were investigated in both positive and negative ion detection modes. The most favourable response was realized in positive ion filament-on mode. After thoroughly purging the system of 1,25(OH)lD3, authentic standards (200 ng each) of vitamins DZ,D,, 25(OH)D3, 1,25(OH),D3 and 24,25(OH)2D3 in mobile phase (50 pL) were introduced separately into the TSP ion source via the column bypass loop. Posditive ion, filament-on TSP mass spectra (mass range mlz 100-600 scanned every 2 s) were recorded for each compound. Quantitative precision of TSP LC/MS: standard curves for vitamin D3. Method (1). Varying amounts (040ng) of vitamin D3 were detected in series by both UV absorbance (264nm) and filament-on TSP MS using HPLC Method B. The [M + HI+ ion (mlz 385) was monitored for TSP detection. Separate standard curves were constructed using the UV peak heights and the TSP peak heights or integrated peak areas. Method (2). In this method an internal standard (40 ng of vitamin D2) was present with each vitamin D3 standard injected and an additional ion channel (mlz 397, [M HI* for vitamin D2)was monitored. Other conditions were exactly as described for method (1). The method of internal standards was used for construction of both UV absorbance and TSP standard curves.
+
In uitro application: la-hydroxylation of 25(OH)D3by guinea pig and sheep renal mitochondria. The preparation of mitochondria, the incubation protocol and initial extraction of the products using CIS solid phase extraction cartridges were performed as described by Hagenfeldt et al. (1988). The crude extracts were dried in a stream of nitrogen and reconstituted in mixture A (50 pL). The reconstituted extracts were analyzed by TSP LC/MS in selected ion monitoring (SIM) mode using HPLC method A. The eluate was monitored at mlz 399 and 402 to detect 1,25(OH),D3 and the added internal standard (96 pmol of [26,26,26-’H3]1 ,25(S)(OH),D3 98 atom YO excess). The amounts of 1,25(OH)2D3formed were estimated from a calibration curve prepared from analysis of standard mixtures. Mixtures containing 96 pmol of [26,26,26-2H3]1,25(S)(OH),D3 (98 atom YO excess) and 096pmol 1,25(OH),D3 were analysed in triplicate by TSP LC/MS in SIM mode. The mass spectrometer was set to monitor mlz399 and 402 with maximized dwell-time and signal averaging within a specified cycle time of 3 s. The ion signals generated by 1,25(OH)2D3and the deuterated analogue from each standard mixture were integrated and linear regression analysis was performed to generate standard curves using the method of internal standards. In uiuo application: vitamin D metabolite prowe in human plasma. Human blood plasma (10 mL) was obtained from a normal adult male volunteer. A lipophilic fraction was obtained by extraction with acetonitrile. This extract was made 50% (vlv) aqueous by addition of dibasic potassium phosphate buffer (pH10.4) and applied to a Waters CIR Sep-Pak cartridge which had been prewashed sequentially with methanol, methano1:water (70:30 vlv) and water ( 5 mL each). After washing with water (2 X 3 mL) and methanol:
ANALYSIS OF VITAMIN D AND ITS METABOLITES USING TSP LClMS [MtHI'
397
;
358
419
428
438
44a
459
388 398 OtHtH1'
488
378
389
399
489
418
429
438
448
458
378
388
398
488
418
429
438
448
458
378
368
1
E
R
?
L
M
e
358
369
E
,
358
369
0
M
,
IMtH-H 01'
Qaa2
1111
WZ
Figure 1. Positive ion filament-assisted TSP mass spectra of vitamin D and metabolites. Each spectrum was obtained from 480 pmol of material; (a) vitamin D,, (b) vitamin D,, (c) 25(OH)D3, (d) 1,25(OH),D3, (el 24,25(OH),D3.
water (70:30 v/v, 2 x 2 mL), vitamin D metabolites were eluted with acetonitri1e:methanol ( 9 5 5 v/v, 2 X 2 mL). This extract, to which was added 96prnol of [26,26,26-2H3]1,25(OH)2D3, was dried under a stream of nitrogen, reconstituted in methano1:water (80:20 v/v, 50 pL) and subjected to TSP LC/MS analysis using HPLC Method C. Ions specific for vitamin D2, vitamin D, and the major metabolites of vitamin D3 were monitored throughout the analysis and a mixture of authentic standards was chromatographed immediately afterwards.
RESULTS AND DISCUSSION This paper describes a new method for the analysis of vitamin D and its metabolites. Several excellent reviews of methodology in this field have appeared during the last ten years (Seamark et af., 1981; Bikle, 1983; Horst, 1985; Porteous et af., 1987). The reliability of many routine clinical assays for these compounds has been shown to be unsatisfactory on the grounds that inter-laboratory variations are unacceptably large Table 1. Retention data for authentic standards using HPLC Methods A-C Compound
Method A
Vitamin Dg
> 60 >60
25-Hydroxyvitamin D3 1,25-DihydroxyvitaminD3 24,25-DihydroxyvitaminD3
26.3 10.4 6.8
Vitamin D2
Retention time lminl Method B Method C
9.8
26.2
10.3
27.0
-
17.4 14.8 12.8
155
(Mayer and Schmidt-Gayk, 1984; Jongen et af., 1984). Combined gas chromatography/mass spectrometry (GUMS) has been advocated for use as a definitive reference technique against which to standardize less specific assays. Stable isotope dilution assays based on GUMS have been developed for most of the important vitamin D3 metabolites (Bjorkhem et al., 1979; Seamark et af., 1980; Bjorkhem and Holmberg, 1980; Coldwell et al., 1984) but few laboratories in the field of vitamin D analysis are equipped to employ these techniques. The development of a clinically useful G U M S assay for 1,25(OH)2D3in blood has proved difficult; the methods reported to date require 10-20mL blood samples and are therefore impractical for routine measurement of normal range concentrations (ca. 0.1 pmol/ mL). In general, published sample preparation procedures for vitamin D and its metabolites prior to G U M S analysis are laborious, typically requiring liquid-liquid extraction followed by two liquid-solid extractions, one or two HPLC purification steps and finally formation of suitable volatile derivatives. Overall recoveries from these lengthy procedures are sometimes rather low, e.g. 34% (Oftebro et d.,1988). A further well-known disadvantage associated with G U M S analysis of these compounds is heat-induced isomerization to pyro and isopyro isomers during gas chromatography (Coldwell et al., 1984). To date this problem has either been tolerated, at the expense of detection limits, or circumvented by pre-analysis isomerization to isotachysterols. We have investigated the TSP LC/MS behaviour of vitamin D and its most important metabolites and evaluated the potential usefulness of the technique as a research tool in this field. TSP mass spectra For initial instrument tuning the vaporizer, ion source block, tip heater and lens temperatures were optimized using background ions ( m / z50 and 318) in positive ion, direct TSP ionization mode. Fine tuning was then performed with the aid of a tuning solution of 1,25(OH)2D3. Operation in positive ion filament-on mode generated substantially increased ion signals compared with direct TSP ionization although the qualitative appearance of the spectrum was unchanged. No appreciable ion currents were observed in negative ion mode. Strong positive ion TSP mass spectra were generated for each compound from 480pmol of material injected (Fig. 1). The mass spectra of vitamins D2 and D3contained only [M HI+ions, while the hydroxylated metabolites additionally formed ammonium adduct ions and fragment ions corresponding to the loss of one or more molecules of water. It is usually difficult to distinguish between positional isomers from TSP mass spectra due to the paucity of fragment ions; however in the case of 1,25(OH)2D, and its 24,25dihydroxy isomer, the striking difference between the relative intensities of the [M + H - H20]+ (mlz 399) and [M+H]+ (mlz 417) signals permitted a reliable distinction to be made. Retention times were subsequently determined for authentic vitamin D and its metabolites using HPLC Method A and by monitoring the appropriate compound specific ions for eluting components. Retention data for all three HPLC methods are shown in Table 1.
+
156
D. WATSON, R. ROSS AND K. D. R.SETCHELL
-------
1,25-d1hydroxrvita~in D3
I
3
1
4
6
Time (minutes) Figure 2. Selected ion monitoring trace showing detection of 0.6 pmol of 1,25(OH),D, by TSP LCIMS.
Detection limit for 1,25(OH),D, We used 1,25(OH)2D3 for fine tuning because this is the most biologically potent metabolite and its concentration in uiuo is normally three orders of magnitude lower than that of vitamin D3 and 25(OH)D3. Hence the detection limit for 1,25(OH)2D3 is an important test of any proposed comprehensive analytical approach for this group of compounds. Under conditions optimized to yield the maximum relative ion current at mlz 399 it was possible to detect 0.6 pmol of 1,25(OH)2D3postLC (SIN 3:1, see Fig. 2).
15 30 Amount (nanograms)
0
Quantitative precision comparison between UV and TSP methods of detection A direct comparison between UV absorbance and TSP detection methods was facilitated by in-series connection of the two detection systems so that a single injection of analyte could be monitored by both. Multiple analyses of each standard mixture were made using each of Methods (1) and (2) described in Experimental Section. A comparison of precision was made by determining the correlation coefficients of linear regression standard curves for UV and TSP
15 30 Amount (nanograms)
0
45
45
(4 0.20
0.20
0
3
2 0.16 d ~
2
5
0.12
.-
0
2 x 2 EL
0.16 0.12
0.0%
0.08
-3 2
0.04
0.04
0.00
9.00 0
5 10 15 Amount (nanograms)
0
5 10 15 Amount (nanograms)
Figure 3. Standard curves for HPLC quantification of vitamin D, using both UV absorbance detection (264 nm) and "in-series" TSP MS detection by Methods (1) and (2). (a) Method (1). TSP detection; (b) Method (1 ), UV detection; (c) Method (21, TSP detection; (d) Method (2). UV detection.
ANALYSIS OF VITAMIN D AND ITS METABOLITES USING TSP LC/MS nlz 385
2555
-
PA25
157
-
.-n Y
ui t u i n
-c m
(b)
3
c
I '
I
'
'
"
19
9
" ' I
'264
I1
dz 397
vitanin Dz
internal standad vitanin
3
l=?=i!= 9
I
5
19
Iine (ninutes)
18 Iine (ninutrs)
9
11
Figure 4. (a) Selected ion monitoring responses for vitamins DZ(mlz397)and D3(mlz385)used for TSP standard curve constructed by Method (2).(b) Ultraviolet absorbance profile (264nm) from the same analytical run as shown in (a). Chromatographic resolution of vitamins D, and D3 was deliberately compromised using method (2) in order to optimize conditions for TSP measurement.
detection in each case. The results of this comparison are shown in Fig. 3. It was strongly apparent that the quantitative .precision of UV detection was superior to that of TSP detection in both cases. This is perhaps surprising for Method (2), where conditions were selected to favour the TSP MS detector by using vitamin D2 as a nearly co-eluting, chemically similar internal standard and exploiting mass specificity to give
absorbance detection. The only further improvement which can be made in the quantitative precision of TSP detection is by the use of the stable isotope internal standard, in which case a direct comparison with UV detection is precluded. An indirect comparison can be made between our stable isotope dilution TSP standard cure for 1,25(OH),D, (Fig. 5 ) and GUMS standard curves for similar isotope dilution assays which have
At the same time conditions for UV measurement were necessarily suboptimal (Fig. 4b). These results clearly demonstrate the intrinsic instability, even over very short time periods, of the ion current signals generated by the TSP ion source and suggest that in general superior quantitative data will be afforded by UV
al., 1979; Watson et al., 1990). These G U M S assays all afford considerably greater quantitative precision than we obtained using TSP mass spectrometric detection in the current study; this further demonstrates the relative weakness of the TSP technique in quantitative applications.
LA-.. A-..-l,.-,.A
L . .
n..m,lr,nr
,.A
-+hown
fR~A-lrh--
nt
LJGGLI U G V G l U p G U UY U U l 3 G l V G J ClllU U L I I b l J [ U J U l h l l G l l l C C
'
I
0,71
*
-
y = 0.0139O8x 0.00437 (r = 0 . 9 4
0.6
a Y
d
O.' -
0 7 0
I
I
10
I
I
I
2Q
, 30
40
AMOUNT (NANOGRAMS) Figure 5. Calibration curve for quantification of 1,25(OH),D3 in the range 0-96 pmol using [26.26,26-*H3]1,25(S)(OH),D3 as internal standard.
D. WATSON, R. ROSS AND K.D. R. SETCHELL
158
0.005
Table 2. Amounts of 1,25-dihydroxyvitamin D3 formed in renal mitochondria1 incubations: quantification by stable isotope dilution and LClMS analysis Extract
Species
1 2 3 4 5 6
Guinea pig Guinea pig Sheep Sheep Sheep Sheep
Incubation time (rninl
ng 1.2510H)~Dsper 100 mg mitochondrialtissue
30
22.0 28.2 23.6 16.7 17.9 23.0
30
15 15 60 60
A 254
0
10
Figure 6. UV absorbance HPLC profile at 264 nm from a simple liquid-solid extract of the supernatant from an incubation of 25(OH)D3 with guinea pig renal mitochondria.
I n vitro application: la-hydroxylation of 25-hydroxyvitamin D3 by guinea pig and sheep renal
mitochondria The renal la-hydroxylation of 25(OH)D3 is the final step in the biogenesis of 1,25(OH),D3, the active antirachitic metabolite of vitamin D3. Extensive mechanis-
tic studies of this metabolic reaction have been carried out in the chicken (Henry and Norman, 1984) and in rachitic rats (Paulson and DeLuca, 1985). In these early animal models relatively large quantities of 1,25(OH)*D3were formed and reliable quantification using competitive binding assays or HPLC with single wavelength UV detection was possible. More recently the latter approach has been extended to measurement of renal la-hydroxylase activity in both pigs (Holmberg et al., 1986) and guinea pigs (Delvin and Dussault, 1985) with normal vitamin D status. Hagenfeldt et al. (1988) developed an isotope dilution method based on GC/MS for quantification of 1,25(OH)*D3produced by guinea pig renal mitochondria in vitro. The UV profiles generated in our experiments from simple liquid-solid extracts (Fig. 6) were complicated by additional intensely absorbing species eluting close to 1,25(OH),D3 and consequently measurement of 1,25(OH)2D3UVresponses in these extracts was difficult. The SIM profiles obtained on-line from the same analyses show far less chemical noise and demonstrate the superior compound specificity obtained with mass spectrometric detection. The SIM responses due to the internal standard and 1,25(OH)J3, formed in the guinea pig mitochondrial incubations were clearly measurable (peaks 111 and IV in Fig. 7). IIC
n/a 399
1.25-dihudm*luituin
r 29529
9
18369
19888.
9588. 2
3
4
5
6
4
5
6
./z 482
s
22438
E
i? -3
B . . . . , 2
I
3
(b) n/z 399
s
111
I - . 9
11
-;
I '
'
I
dWU@+'' 9
,
. . . ,
5
18
,
. , 15
.
,
,
20
.
.
.
~
25
,
,
, : , 38
line (Minutes) IlnE (IIIHu11)
Figure 7. Selected ion monitoring profiles from: (a) A standard mixture containing 40 ng ezch of 1,25(OH)zD, and [26,26,26-2H3]1,25(S)(OH)2D3.(b) The same incubation extract which generated the UV absorbance profile shown in Fig. 6.
Figure 8. TSP LClMS summed ion chromatogram (summed SIM signals at mlr 399, 401, 402 and 417) of human plasma extract showing detection of vitamin D, and its principal metabolites using HPLC Method C: peak l, 24,25(OH)?D3; peak II, 1,25(OH)2D3 plus trideuterated internal standard; peak 111, 25(OH)D3; peak IV, vitamin D,.
ANALYSIS OF VITAMIN D AND ITS METABOLITES USING TSP LClMS
In the experiments using sheep mitochondria a modified mobile phase was used (methanol:water, 80:20, v/v containing 2% formic acid) in order to determine whether the intensity of the [M + H - H,O]+ signal for 1,25(OH),D3could be increased by eliminating ammonium adduct formation. Observed signal-to-noise ratios were clearly lower in this experiment, indicating that 2% formic acid is less efficient for primary ionization of 1,25(OH)2D3 than 0.1 M ammonium acetate. The standard curve obtained from SIM analyses for 1,25(OH)2D3using [26,26,26-*H,]1,25(S)(OH),D, as internal standard is shown in Fig. 5 and the measured amounts of 1,25(OH)2D3formed in incubations using guinea pig and sheep mitochondria are listed in Table 2. In addition to 1,25(OH)2D3,the m/z 399 ion current profiles from some incubation extracts from both guinea pig and sheep mitochondria contained responses arising from two other, more hydrophilic species (peaks I and I1 in Fig. 7). The 24,25- and 25,26dihydroxyvitamin D3isomers are known to elute earlier than 1,25(OH)2D3in reversed phase systems and the presence of these metabolites in the incubation extracts probably accounts for peaks I and I1 in the m / z 399 ion profiles. Authentic 24,25(OH),D, had a mass spectrum and elution time relative to [ *H3]1,25(OH),D, identical to those of peak 11.
In uivo application: vitamin D metabolite profile in human plasma The UV absorbance profile resulting from gradient elution of the plasma extract was extremely complex and it was not possible to confidently assign specific responses to vitamin D metabolites. From the TSP SIM profiles, however, it was possible to detect vitamin D3, the added deuterated standard and the three principal metabolites of vitamin D3 and to generate the reconstructed total ion current profile shown in Fig. 8. The identities of peaks I-IV were confirmed in terms of
159
retention time and compound-specific mass using a standard mixture chromatographed immediately after the plasma extract. Considerable contamination of the HPLC column was caused by the plasma extract, as evidenced by a high UV absorbance background throughout the standard mixture analysis. The contamination was removed after several blank runs where methanol only was injected. TSP LC/MS clearly has strong potential for multicomponent analysis of vitamin D metabolites in blood plasma, although to obtain quantitative data, particularly from a gradient elution, stable isotopically labelled standards for each component will be required, due to the compound and solvent composition dependent response of the TSP ionization process.
CONCLUSION
In conclusion, our results indicate that TSP LC/MS is useful for the detection of vitamin D and its major biologically active metabolites and can be applied to in v i m and in uivo studies of metabolism of these compounds. The sample preparation procedures required prior to TSP LC/MS analysis are simpler than those required for G U M S analysis while compound specificity is not compromised, although the precision of quantitative measurements compares rather poorly with other techniques. We anticipate that TSP LC/MS will play an increasingly important role, complementary to that of GUMS, in many aspects of vitamin D research in the future.
Acknowledgements Professor I. Bjorkhem (Huddinge Hospital, Stockholm, Sweden) is thanked for supplying the guinea pig renal mitochondria1 incubation extracts used in this work.
REFERENCES Asknes, L. (1980). Clin. Chim. Acta 104, 147. Belsey, R., De Luca, H. F. and Potts, J. T. (1971).J. Clin. Endocrin. Metabol. 33, 554. Bikle, D. 0. (1983). Assay of Calcium-regufafing Hormones. Springer-Verlag, New York. Bjorkhem, 1. and Holmberg, 1. (1980). Meth. fnzym. 67, 385. Bjorkhem, I., Holmberg, I., Kristiansen, T. and Pedersen, J. I. (1979). Clin. Chem. 25, 584. Blakely, C. R. and Vestal, M. L. (1983). Anal. Chem. 55, 750. Chandlsr, J. S., Pike, J. W., Hagan, L. A. and Haussler, M. R. (1980). Mefh. fnzym. 67, 522. Clemens, T. L., Adams, J. S., Nolan, J. H. and Holick, M. F. (1982). Clin. Chim. Acta 121, 301. Coldwell, R. D., Trafford, 0 . J. H., Makin, H. L. J., Varley, M. J. and Kirk, 0. N. (1984). Clin. Chem. 30, 1193. Covey, R. T., Crowther, J. B., Dewey, E. A. and Henion, J. D. (1985). Anal. Chem. 57,474. Delvin, E. E. and Dussault, M. (1985). Arch. Biochem. Biophys. 240,337. Gaskell, S. J., Rollins, K., Smith, R. W. and Parker, C. E. (1987). Biomed. Environ. Mass Spectrom. 14, 717. Gray, T. K., Mcadoo, T., Pool, D., Lester, G. E. and Williams, M. E. (1981). Clin. Chem. 27, 458. Hagenfeldt, Y., Pedersen, J. I. and Bjorkhem, I. (1988). Biochem. J. 250, 521.
Henry, H. 1. and Norman, A. W.(1984). Ann. Rev. Nutr. 4,493. Hollis, B. W. and Napoli, J. L. (1985). Clin. Chem. 31, 1815. Holmberg, I . , Saarem, K., Pedersen, J. 1. and Bjorkhem, 1. (1986). Anal. Biochem. 159,317. Horst, R. L. (1985). In Vitamin D: Basic and Clinical Aspects, ed. by Kumar, R., p. 423. Martinus Nijhoff, Boston. Horst, R. L., Reinhardt, T. A., Beitr, D. C. and Littledike, E. T. (1981). Steroids 37, 581. Jongen, M. J. M., Van der Vijgh, W. J. F., Willems, H. J. J. and Netelenbos, J. C. (1981). Clin. Chem. 27, 444. Jongen, M. J. M., Van Ginkel, F. C., Van der Vijgh, W. J. F., Kuiper, S., Netelenbos, J. C. and Lips P. (1984). Clin. Chem. 30. .., 399. .._ Kim, H. Y., Yergey, J. A. and Salem, N. Jr. (1987). J. Chromafogr. 394,155. Mayer. E. and Schmidt-Gayk, H. (1984). Clin. Chem. 30,1199. Oftenbro, H., Falch, J. A., Holmberg, I. and Haug, E. (1988). Clin. Chim. Acta 176, 157. Okano, T., Mizuno, N., Shida. S., Takahashi, N., Kobayashi, T., Kuroda, E., Kodama, S. and Matsuo, T. (1981). J. Nutr. Sci. Vitamin 27, 43. Parker, C. E., de Wit, J. S. M., Smith, R. W., Gopinathan, M. B., Hernandez, 0..Tomer, K. 6..Vestal, C. H., Sanders, J. M . and Bend, J. R. (1988). Biomed. Environ. Mass Spectrom. 15, 623.
160
D. WATSON, R . ROSS AND K.D. R.SETCHELL
Paulson, S. K. and DeLoca, H. F. (1985). J. Biof. Chem. 260, 11488. Porteous, C. E., Coldwell, R. D., Trafford, D. J. H. and Makin, H. L. J. (1987). J. Steroid Biochern. 28, 785. Richmond, R., Clarke, S. R., Watson, D., Chappell, C. G.,Dollery, C. T. and Taylor, G. W. (1986). Biochim. Biophys. Acta 881, 159. Seamark, D.A., Trafford, D.J. H. and Makin, H. L.J.(1980). Clin, Chim. Acta 106, 51. Seamark, D. A., Trafford, D. J. H. and Makin, H. L. J. (1981). J. Steroid Biochem. 14, 111. Setchell, K. D. R. and Vestal, C. H. (1989). J. Lipid Res. Setchell, K. D. R., Welsh, M. 6. and Lim, C. K. (1987). J. Chromatogr. 386. 315.
Voyksner, R. D. and Bush, E. D.(1987). Biomed. Environ. Mass Spectrom. 14, 213. Watson, D., Taylor, G.W. and Murray, S. (1985). Biorned. Mass Spectrorn. 12, 610. Watson, D.. Taylor, G. W. and Murray, S. (1986). Biomed. Environ. Mass Spectrom. 13, 65. Watson, D., Taylor, G. W., Laird, S. and Vinson, G. P. (1987). Biochem. J. 242, 109. Watson, D., Taylor, G. W., Wilson, R., Cole, P. J. and Rowe, C. (1988). Biomed. Environ. Mass Spectrom. 17, 251. Watson, D., Ross R. and Setchell, K. D. R. (1990). Proceedings of the 38th ASMS Meeting on Mass Spectrometry and Allied Topics.
Analysis of Vitamin D and its Metabolites using Thermospray Liquid Chromatography/Mass Spectrometry David Watson and Kenneth D. R. Setchell* Mass Spectrometry Laboratory, Children’s Hospital Medical Center, Elland and Bethesda Avenues. Cincinnati, Ohio 45229, USA
Richardus Ross Department of Neonatology, Children’s Hospital Medical Center, Elland and Bethesda Avenues, Cincinnati, Ohio 45229, USA
A new method is described for the analysis of vitamin D and its metabolites utilizing thermospray (TSP) mass spectrometry as an on-line detector for high performance liquid chromatography. Ionization conditions were optimized for use with isocratic reversed phase chromatography. TSP mass spectrometry was employed in series with a W absorbance detector to facilitate comparisons between the two methods of detection. Positive ion TSP mass spectra were recorded for vitamin D2, vitamin D3, 25-hydroxyvitamin D3 (25(0H)D3), 1,25dihydroxyvitamin D, (1,25(OH),D3) and 24,25-dihydroxyvitamin D3 (24,25(OH),D,). The spectra contained protonated molecular ions, ammonium adduct ions and fragment ions due to the loss of one or more molecules of water. A comparison of quantitative precision was made by determining UV absorbance and TSP standard curves for vitamin D, using two different methods: (1)External standard method with post-column (post UV detector) addition of ammonium acetate. (2) As (1) but using the method of internal standards with a closely eluting internal standard (vitamin Dz). In each case the quantitative precision (correlation coefficient) for UV absorbance detection was superior owing to intrinsic instability of the TSP ion beam. A stable isotopically labelled internal standard was employed in the development of an assay for 1,25(OH),D,. The assay was used to quantify in uitro enzymic conversion of 25(OH)D3 to 1,25(oH),D3 in guinea pig and sheep renal mitochondria1 incubations. TSP LC/MS was also applied to analysis of an extract of human blood plasma in which D3and each of its principal metabolites were identified in a single analysis.
INTRODUCTION The metabolism of vitamin D and the biological activities of vitamin D metabolites have been studied with growing interest during the past 25 years. In order to facilitate clinical and biochemical investigations much effort has focussed on the development of appropriate analytical methods. Currently the methods routinely used for quantification of vitamin D3 metabolites include competitive protein binding assays (Belsey et al., 1971; Horst et al., 198l), radioreceptor assays (Chandler et al., 1980; Oftebro et af., 1988), radioimmonoassays (Gray et al., 1981; Hollis and Napoli, 1985) and high performance liquid chromatography (HPLC) with ultraviolet (UV) absorbance detection (Clemens et al., 1982; Okano et al., 1981). HPLC has also been employed in conjunction with competitive binding assays (Asknes, 1980; Jongen et al., 1981) to furnish greater compound specificity and to facilitate multicomponent analysis. As a result of the commercialization of the thermospray (TSP) interface and ion source (Blakley and Vestal, 1983) it is now routinely possible to introduce the entire eluate from analytical scale HPLC systems into a mass spectrometer without loss of chromato* Author to whom correspondence should be addressed. 0269-3879/91/040153-08 $05.00 01991 by John Wiley & Sons, Ltd.
graphic resolution, and on-line mass spectrometric analysis of polar and thermally labile compounds eluting from the HPLC column can be readily accomplished. Biomedical applications of TSP liquid chromatography/mass spectrometry (LC/MS) are now numerous, including analyses of drugs and drug metabolites (Covey et af., 1985), steroid conjugates (Watson et nl., 1985, 1986) corticosteroid hormones (Watson et al., 1987; Gaskell et al., 1987), eicosanoids (Richmond et al., 1986; Voyksner and Bush, 1987), phospholipids (Kim et al., 1987), glutathione conjugates (Parker et al., 1988), phytoestrogens (Setchell et al., 1987), bacterial phenazines (Watson et af., 1988), bile acids (Setchell and Vestal, 1989) and many others. Since HPLC is by far the most popular chromatographic technique in use for the purification and assay of vitamin D and its metabolites there is much potential benefit to be gained from the application of TSP mass spectrometry as an HPLC detector for these compounds. In concept TSP LC/MS possesses characteristics which should afford several advantages for analysis of vitamin D and its metabolites compared with gas chromatography/mass spectrometry (GUMS); it should be possible to reduce the number of sample extraction steps, no derivatization is required prior to chromatography and thermal effects should be eliminated. For these reasons we have investigated the qualitative and quantitative aspects of the TSP mass spectrometric behaviour of the most biologically important vitamin D related compounds. Received 21 August I990 Accepted I1 September I990
154
D. WATSON, R. ROSS AND K. D. R. SETCHELL
EXPERIMENTAL Chemicals. Vitamins D2 and D3 were purchased from Sigma Chemical Co. (St Louis, MO, USA). Vitamin D metabolites were generously donated by Hoffmann-la-Roche, Nutley, NJ, USA. These materials were stored in ethanol at -20 “C when not in use and the concentrations of stock solutions were verified by UV absorbance prior to use in quantitative analyses. Water for HPLC was purified to 18 M B cm resistivity; HPLC quality methanol was purchased from Baxter Healthcare Corporation, Muskegon, MI, USA. Ammonium acetate (HPLC grade) was obtained from J. T. Baker Chemical Co., Phillipsburg, NJ, USA. All other chemicals and biochemicals were standard commercial high purity preparations. Instrumentation. The HPLC mobile phase was delivered by a Waters Model 600-MS pump and the eluate was passed through the high pressure flow-cell of a Waters Model 490MS variable wavelength detector (set to monitor at 264 nm) connected directly to the vaporizer tube of a Vestec 201 TSP LClMS system (Vestec Corp., Houston, TX, USA). This system has a single quadrupole mass analyser (10-800 Da, unit resolution). All of the analyses were performed using a standard Vestec vaporizer tube with a 150 pm spray orifice. Samples were introduced into the mass spectrometer either via an HPLC column or using a column bypass loop with a Waters Model U6K injector. Instrument control and data acquisition were performed using a Teknivent Vector 1 (Teknivent, St Louis, MO, USA) mass spectrometer data system. High performance liquid chromatography. Three different reversed phase chromatography systems were developed in the course of these studies; while each was designed partly to satisfy the operating requirements of the TSP ionization process, the elution conditions are compatible with those used generally in the field of vitamin D research: Method A. For separation of hydroxylated vitamin D metabolites a Waters Nova-PAK c 1 8 column (15 cm X 0.46 cm i.d., 4 pm particle size) was eluted isocratically with methano1:water (80:20 v/v) containing ammonium acetate (0.1 M) at a flow rate of 1 mL/min. Method B. For Dz and D3 analysis a Hypersil CIRcolumn (25 cm X 0.46 cm i.d., 5 pm particle size, Keystone Scientific Inc., Bellefonte, PA, USA) was eluted with methanol (0.9 mL/min). The ionizing buffer for the TSP source (0.4 M ammonium acetate, pH 6.5, flow rate 0.3 mL/min) was added via a Valco zero dead volume “tee” connector on the outlet side of the UV absorbance detector using a separate pulsedampled HPLC pump (Waters Model 590). Method C. For simultaneous analysis of vitamins D2, D3 and their hydroxylated metabolites a combination of Methods A and B were used; the Hypersil column was eluted at 0.9 mllmin with a linear solvent gradient running from methanokwater (80:20 vlv) to 100% methanol over a 15-min period. The final solvent composition was held for 10 min at the end of the gradient. The ionizing buffer was added as in Method B. TSP mass spectra. The operating parameters of the TSP LC/MS system (vaporizer control temperature, vaporizer tip temperature, ion source block temperature) were initially optimized using background ions generated by ionization of HPLC Method A mobile phase which was continuously pumped into the ion source at a flow rate of lmL/min.
Ionization was enhanced when the ion source filament was switched on. Detailed optimization of ionization was then carried out using a “tuning solution” (a 2 pg/mL solution of 1,25(OH)2D3in the mobile phase, pumped continuously via the column bypass loop into the ion source at a flowrate of lmL/min). All available modes of ionization (pure TSP, “filament-ion’’ and “discharge-on”) were investigated in both positive and negative ion detection modes. The most favourable response was realized in positive ion filament-on mode. After thoroughly purging the system of 1,25(OH)lD3, authentic standards (200 ng each) of vitamins DZ,D,, 25(OH)D3, 1,25(OH),D3 and 24,25(OH)2D3 in mobile phase (50 pL) were introduced separately into the TSP ion source via the column bypass loop. Posditive ion, filament-on TSP mass spectra (mass range mlz 100-600 scanned every 2 s) were recorded for each compound. Quantitative precision of TSP LC/MS: standard curves for vitamin D3. Method (1). Varying amounts (040ng) of vitamin D3 were detected in series by both UV absorbance (264nm) and filament-on TSP MS using HPLC Method B. The [M + HI+ ion (mlz 385) was monitored for TSP detection. Separate standard curves were constructed using the UV peak heights and the TSP peak heights or integrated peak areas. Method (2). In this method an internal standard (40 ng of vitamin D2) was present with each vitamin D3 standard injected and an additional ion channel (mlz 397, [M HI* for vitamin D2)was monitored. Other conditions were exactly as described for method (1). The method of internal standards was used for construction of both UV absorbance and TSP standard curves.
+
In uitro application: la-hydroxylation of 25(OH)D3by guinea pig and sheep renal mitochondria. The preparation of mitochondria, the incubation protocol and initial extraction of the products using CIS solid phase extraction cartridges were performed as described by Hagenfeldt et al. (1988). The crude extracts were dried in a stream of nitrogen and reconstituted in mixture A (50 pL). The reconstituted extracts were analyzed by TSP LC/MS in selected ion monitoring (SIM) mode using HPLC method A. The eluate was monitored at mlz 399 and 402 to detect 1,25(OH),D3 and the added internal standard (96 pmol of [26,26,26-’H3]1 ,25(S)(OH),D3 98 atom YO excess). The amounts of 1,25(OH)2D3formed were estimated from a calibration curve prepared from analysis of standard mixtures. Mixtures containing 96 pmol of [26,26,26-2H3]1,25(S)(OH),D3 (98 atom YO excess) and 096pmol 1,25(OH),D3 were analysed in triplicate by TSP LC/MS in SIM mode. The mass spectrometer was set to monitor mlz399 and 402 with maximized dwell-time and signal averaging within a specified cycle time of 3 s. The ion signals generated by 1,25(OH)2D3and the deuterated analogue from each standard mixture were integrated and linear regression analysis was performed to generate standard curves using the method of internal standards. In uiuo application: vitamin D metabolite prowe in human plasma. Human blood plasma (10 mL) was obtained from a normal adult male volunteer. A lipophilic fraction was obtained by extraction with acetonitrile. This extract was made 50% (vlv) aqueous by addition of dibasic potassium phosphate buffer (pH10.4) and applied to a Waters CIR Sep-Pak cartridge which had been prewashed sequentially with methanol, methano1:water (70:30 vlv) and water ( 5 mL each). After washing with water (2 X 3 mL) and methanol:
ANALYSIS OF VITAMIN D AND ITS METABOLITES USING TSP LClMS [MtHI'
397
;
358
419
428
438
44a
459
388 398 OtHtH1'
488
378
389
399
489
418
429
438
448
458
378
388
398
488
418
429
438
448
458
378
368
1
E
R
?
L
M
e
358
369
E
,
358
369
0
M
,
IMtH-H 01'
Qaa2
1111
WZ
Figure 1. Positive ion filament-assisted TSP mass spectra of vitamin D and metabolites. Each spectrum was obtained from 480 pmol of material; (a) vitamin D,, (b) vitamin D,, (c) 25(OH)D3, (d) 1,25(OH),D3, (el 24,25(OH),D3.
water (70:30 v/v, 2 x 2 mL), vitamin D metabolites were eluted with acetonitri1e:methanol ( 9 5 5 v/v, 2 X 2 mL). This extract, to which was added 96prnol of [26,26,26-2H3]1,25(OH)2D3, was dried under a stream of nitrogen, reconstituted in methano1:water (80:20 v/v, 50 pL) and subjected to TSP LC/MS analysis using HPLC Method C. Ions specific for vitamin D2, vitamin D, and the major metabolites of vitamin D3 were monitored throughout the analysis and a mixture of authentic standards was chromatographed immediately afterwards.
RESULTS AND DISCUSSION This paper describes a new method for the analysis of vitamin D and its metabolites. Several excellent reviews of methodology in this field have appeared during the last ten years (Seamark et af., 1981; Bikle, 1983; Horst, 1985; Porteous et af., 1987). The reliability of many routine clinical assays for these compounds has been shown to be unsatisfactory on the grounds that inter-laboratory variations are unacceptably large Table 1. Retention data for authentic standards using HPLC Methods A-C Compound
Method A
Vitamin Dg
> 60 >60
25-Hydroxyvitamin D3 1,25-DihydroxyvitaminD3 24,25-DihydroxyvitaminD3
26.3 10.4 6.8
Vitamin D2
Retention time lminl Method B Method C
9.8
26.2
10.3
27.0
-
17.4 14.8 12.8
155
(Mayer and Schmidt-Gayk, 1984; Jongen et af., 1984). Combined gas chromatography/mass spectrometry (GUMS) has been advocated for use as a definitive reference technique against which to standardize less specific assays. Stable isotope dilution assays based on GUMS have been developed for most of the important vitamin D3 metabolites (Bjorkhem et al., 1979; Seamark et af., 1980; Bjorkhem and Holmberg, 1980; Coldwell et al., 1984) but few laboratories in the field of vitamin D analysis are equipped to employ these techniques. The development of a clinically useful G U M S assay for 1,25(OH)2D3in blood has proved difficult; the methods reported to date require 10-20mL blood samples and are therefore impractical for routine measurement of normal range concentrations (ca. 0.1 pmol/ mL). In general, published sample preparation procedures for vitamin D and its metabolites prior to G U M S analysis are laborious, typically requiring liquid-liquid extraction followed by two liquid-solid extractions, one or two HPLC purification steps and finally formation of suitable volatile derivatives. Overall recoveries from these lengthy procedures are sometimes rather low, e.g. 34% (Oftebro et d.,1988). A further well-known disadvantage associated with G U M S analysis of these compounds is heat-induced isomerization to pyro and isopyro isomers during gas chromatography (Coldwell et al., 1984). To date this problem has either been tolerated, at the expense of detection limits, or circumvented by pre-analysis isomerization to isotachysterols. We have investigated the TSP LC/MS behaviour of vitamin D and its most important metabolites and evaluated the potential usefulness of the technique as a research tool in this field. TSP mass spectra For initial instrument tuning the vaporizer, ion source block, tip heater and lens temperatures were optimized using background ions ( m / z50 and 318) in positive ion, direct TSP ionization mode. Fine tuning was then performed with the aid of a tuning solution of 1,25(OH)2D3. Operation in positive ion filament-on mode generated substantially increased ion signals compared with direct TSP ionization although the qualitative appearance of the spectrum was unchanged. No appreciable ion currents were observed in negative ion mode. Strong positive ion TSP mass spectra were generated for each compound from 480pmol of material injected (Fig. 1). The mass spectra of vitamins D2 and D3contained only [M HI+ions, while the hydroxylated metabolites additionally formed ammonium adduct ions and fragment ions corresponding to the loss of one or more molecules of water. It is usually difficult to distinguish between positional isomers from TSP mass spectra due to the paucity of fragment ions; however in the case of 1,25(OH)2D, and its 24,25dihydroxy isomer, the striking difference between the relative intensities of the [M + H - H20]+ (mlz 399) and [M+H]+ (mlz 417) signals permitted a reliable distinction to be made. Retention times were subsequently determined for authentic vitamin D and its metabolites using HPLC Method A and by monitoring the appropriate compound specific ions for eluting components. Retention data for all three HPLC methods are shown in Table 1.
+
156
D. WATSON, R. ROSS AND K. D. R.SETCHELL
-------
1,25-d1hydroxrvita~in D3
I
3
1
4
6
Time (minutes) Figure 2. Selected ion monitoring trace showing detection of 0.6 pmol of 1,25(OH),D, by TSP LCIMS.
Detection limit for 1,25(OH),D, We used 1,25(OH)2D3 for fine tuning because this is the most biologically potent metabolite and its concentration in uiuo is normally three orders of magnitude lower than that of vitamin D3 and 25(OH)D3. Hence the detection limit for 1,25(OH)2D3 is an important test of any proposed comprehensive analytical approach for this group of compounds. Under conditions optimized to yield the maximum relative ion current at mlz 399 it was possible to detect 0.6 pmol of 1,25(OH)2D3postLC (SIN 3:1, see Fig. 2).
15 30 Amount (nanograms)
0
Quantitative precision comparison between UV and TSP methods of detection A direct comparison between UV absorbance and TSP detection methods was facilitated by in-series connection of the two detection systems so that a single injection of analyte could be monitored by both. Multiple analyses of each standard mixture were made using each of Methods (1) and (2) described in Experimental Section. A comparison of precision was made by determining the correlation coefficients of linear regression standard curves for UV and TSP
15 30 Amount (nanograms)
0
45
45
(4 0.20
0.20
0
3
2 0.16 d ~
2
5
0.12
.-
0
2 x 2 EL
0.16 0.12
0.0%
0.08
-3 2
0.04
0.04
0.00
9.00 0
5 10 15 Amount (nanograms)
0
5 10 15 Amount (nanograms)
Figure 3. Standard curves for HPLC quantification of vitamin D, using both UV absorbance detection (264 nm) and "in-series" TSP MS detection by Methods (1) and (2). (a) Method (1). TSP detection; (b) Method (1 ), UV detection; (c) Method (21, TSP detection; (d) Method (2). UV detection.
ANALYSIS OF VITAMIN D AND ITS METABOLITES USING TSP LC/MS nlz 385
2555
-
PA25
157
-
.-n Y
ui t u i n
-c m
(b)
3
c
I '
I
'
'
"
19
9
" ' I
'264
I1
dz 397
vitanin Dz
internal standad vitanin
3
l=?=i!= 9
I
5
19
Iine (ninutes)
18 Iine (ninutrs)
9
11
Figure 4. (a) Selected ion monitoring responses for vitamins DZ(mlz397)and D3(mlz385)used for TSP standard curve constructed by Method (2).(b) Ultraviolet absorbance profile (264nm) from the same analytical run as shown in (a). Chromatographic resolution of vitamins D, and D3 was deliberately compromised using method (2) in order to optimize conditions for TSP measurement.
detection in each case. The results of this comparison are shown in Fig. 3. It was strongly apparent that the quantitative .precision of UV detection was superior to that of TSP detection in both cases. This is perhaps surprising for Method (2), where conditions were selected to favour the TSP MS detector by using vitamin D2 as a nearly co-eluting, chemically similar internal standard and exploiting mass specificity to give
absorbance detection. The only further improvement which can be made in the quantitative precision of TSP detection is by the use of the stable isotope internal standard, in which case a direct comparison with UV detection is precluded. An indirect comparison can be made between our stable isotope dilution TSP standard cure for 1,25(OH),D, (Fig. 5 ) and GUMS standard curves for similar isotope dilution assays which have
At the same time conditions for UV measurement were necessarily suboptimal (Fig. 4b). These results clearly demonstrate the intrinsic instability, even over very short time periods, of the ion current signals generated by the TSP ion source and suggest that in general superior quantitative data will be afforded by UV
al., 1979; Watson et al., 1990). These G U M S assays all afford considerably greater quantitative precision than we obtained using TSP mass spectrometric detection in the current study; this further demonstrates the relative weakness of the TSP technique in quantitative applications.
LA-.. A-..-l,.-,.A
L . .
n..m,lr,nr
,.A
-+hown
fR~A-lrh--
nt
LJGGLI U G V G l U p G U UY U U l 3 G l V G J ClllU U L I I b l J [ U J U l h l l G l l l C C
'
I
0,71
*
-
y = 0.0139O8x 0.00437 (r = 0 . 9 4
0.6
a Y
d
O.' -
0 7 0
I
I
10
I
I
I
2Q
, 30
40
AMOUNT (NANOGRAMS) Figure 5. Calibration curve for quantification of 1,25(OH),D3 in the range 0-96 pmol using [26.26,26-*H3]1,25(S)(OH),D3 as internal standard.
D. WATSON, R. ROSS AND K.D. R. SETCHELL
158
0.005
Table 2. Amounts of 1,25-dihydroxyvitamin D3 formed in renal mitochondria1 incubations: quantification by stable isotope dilution and LClMS analysis Extract
Species
1 2 3 4 5 6
Guinea pig Guinea pig Sheep Sheep Sheep Sheep
Incubation time (rninl
ng 1.2510H)~Dsper 100 mg mitochondrialtissue
30
22.0 28.2 23.6 16.7 17.9 23.0
30
15 15 60 60
A 254
0
10
Figure 6. UV absorbance HPLC profile at 264 nm from a simple liquid-solid extract of the supernatant from an incubation of 25(OH)D3 with guinea pig renal mitochondria.
I n vitro application: la-hydroxylation of 25-hydroxyvitamin D3 by guinea pig and sheep renal
mitochondria The renal la-hydroxylation of 25(OH)D3 is the final step in the biogenesis of 1,25(OH),D3, the active antirachitic metabolite of vitamin D3. Extensive mechanis-
tic studies of this metabolic reaction have been carried out in the chicken (Henry and Norman, 1984) and in rachitic rats (Paulson and DeLuca, 1985). In these early animal models relatively large quantities of 1,25(OH)*D3were formed and reliable quantification using competitive binding assays or HPLC with single wavelength UV detection was possible. More recently the latter approach has been extended to measurement of renal la-hydroxylase activity in both pigs (Holmberg et al., 1986) and guinea pigs (Delvin and Dussault, 1985) with normal vitamin D status. Hagenfeldt et al. (1988) developed an isotope dilution method based on GC/MS for quantification of 1,25(OH)*D3produced by guinea pig renal mitochondria in vitro. The UV profiles generated in our experiments from simple liquid-solid extracts (Fig. 6) were complicated by additional intensely absorbing species eluting close to 1,25(OH),D3 and consequently measurement of 1,25(OH)2D3UVresponses in these extracts was difficult. The SIM profiles obtained on-line from the same analyses show far less chemical noise and demonstrate the superior compound specificity obtained with mass spectrometric detection. The SIM responses due to the internal standard and 1,25(OH)J3, formed in the guinea pig mitochondrial incubations were clearly measurable (peaks 111 and IV in Fig. 7). IIC
n/a 399
1.25-dihudm*luituin
r 29529
9
18369
19888.
9588. 2
3
4
5
6
4
5
6
./z 482
s
22438
E
i? -3
B . . . . , 2
I
3
(b) n/z 399
s
111
I - . 9
11
-;
I '
'
I
dWU@+'' 9
,
. . . ,
5
18
,
. , 15
.
,
,
20
.
.
.
~
25
,
,
, : , 38
line (Minutes) IlnE (IIIHu11)
Figure 7. Selected ion monitoring profiles from: (a) A standard mixture containing 40 ng ezch of 1,25(OH)zD, and [26,26,26-2H3]1,25(S)(OH)2D3.(b) The same incubation extract which generated the UV absorbance profile shown in Fig. 6.
Figure 8. TSP LClMS summed ion chromatogram (summed SIM signals at mlr 399, 401, 402 and 417) of human plasma extract showing detection of vitamin D, and its principal metabolites using HPLC Method C: peak l, 24,25(OH)?D3; peak II, 1,25(OH)2D3 plus trideuterated internal standard; peak 111, 25(OH)D3; peak IV, vitamin D,.
ANALYSIS OF VITAMIN D AND ITS METABOLITES USING TSP LClMS
In the experiments using sheep mitochondria a modified mobile phase was used (methanol:water, 80:20, v/v containing 2% formic acid) in order to determine whether the intensity of the [M + H - H,O]+ signal for 1,25(OH),D3could be increased by eliminating ammonium adduct formation. Observed signal-to-noise ratios were clearly lower in this experiment, indicating that 2% formic acid is less efficient for primary ionization of 1,25(OH)2D3 than 0.1 M ammonium acetate. The standard curve obtained from SIM analyses for 1,25(OH)2D3using [26,26,26-*H,]1,25(S)(OH),D, as internal standard is shown in Fig. 5 and the measured amounts of 1,25(OH)2D3formed in incubations using guinea pig and sheep mitochondria are listed in Table 2. In addition to 1,25(OH)2D3,the m/z 399 ion current profiles from some incubation extracts from both guinea pig and sheep mitochondria contained responses arising from two other, more hydrophilic species (peaks I and I1 in Fig. 7). The 24,25- and 25,26dihydroxyvitamin D3isomers are known to elute earlier than 1,25(OH)2D3in reversed phase systems and the presence of these metabolites in the incubation extracts probably accounts for peaks I and I1 in the m / z 399 ion profiles. Authentic 24,25(OH),D, had a mass spectrum and elution time relative to [ *H3]1,25(OH),D, identical to those of peak 11.
In uivo application: vitamin D metabolite profile in human plasma The UV absorbance profile resulting from gradient elution of the plasma extract was extremely complex and it was not possible to confidently assign specific responses to vitamin D metabolites. From the TSP SIM profiles, however, it was possible to detect vitamin D3, the added deuterated standard and the three principal metabolites of vitamin D3 and to generate the reconstructed total ion current profile shown in Fig. 8. The identities of peaks I-IV were confirmed in terms of
159
retention time and compound-specific mass using a standard mixture chromatographed immediately after the plasma extract. Considerable contamination of the HPLC column was caused by the plasma extract, as evidenced by a high UV absorbance background throughout the standard mixture analysis. The contamination was removed after several blank runs where methanol only was injected. TSP LC/MS clearly has strong potential for multicomponent analysis of vitamin D metabolites in blood plasma, although to obtain quantitative data, particularly from a gradient elution, stable isotopically labelled standards for each component will be required, due to the compound and solvent composition dependent response of the TSP ionization process.
CONCLUSION
In conclusion, our results indicate that TSP LC/MS is useful for the detection of vitamin D and its major biologically active metabolites and can be applied to in v i m and in uivo studies of metabolism of these compounds. The sample preparation procedures required prior to TSP LC/MS analysis are simpler than those required for G U M S analysis while compound specificity is not compromised, although the precision of quantitative measurements compares rather poorly with other techniques. We anticipate that TSP LC/MS will play an increasingly important role, complementary to that of GUMS, in many aspects of vitamin D research in the future.
Acknowledgements Professor I. Bjorkhem (Huddinge Hospital, Stockholm, Sweden) is thanked for supplying the guinea pig renal mitochondria1 incubation extracts used in this work.
REFERENCES Asknes, L. (1980). Clin. Chim. Acta 104, 147. Belsey, R., De Luca, H. F. and Potts, J. T. (1971).J. Clin. Endocrin. Metabol. 33, 554. Bikle, D. 0. (1983). Assay of Calcium-regufafing Hormones. Springer-Verlag, New York. Bjorkhem, 1. and Holmberg, 1. (1980). Meth. fnzym. 67, 385. Bjorkhem, I., Holmberg, I., Kristiansen, T. and Pedersen, J. I. (1979). Clin. Chem. 25, 584. Blakely, C. R. and Vestal, M. L. (1983). Anal. Chem. 55, 750. Chandlsr, J. S., Pike, J. W., Hagan, L. A. and Haussler, M. R. (1980). Mefh. fnzym. 67, 522. Clemens, T. L., Adams, J. S., Nolan, J. H. and Holick, M. F. (1982). Clin. Chim. Acta 121, 301. Coldwell, R. D., Trafford, 0 . J. H., Makin, H. L. J., Varley, M. J. and Kirk, 0. N. (1984). Clin. Chem. 30, 1193. Covey, R. T., Crowther, J. B., Dewey, E. A. and Henion, J. D. (1985). Anal. Chem. 57,474. Delvin, E. E. and Dussault, M. (1985). Arch. Biochem. Biophys. 240,337. Gaskell, S. J., Rollins, K., Smith, R. W. and Parker, C. E. (1987). Biomed. Environ. Mass Spectrom. 14, 717. Gray, T. K., Mcadoo, T., Pool, D., Lester, G. E. and Williams, M. E. (1981). Clin. Chem. 27, 458. Hagenfeldt, Y., Pedersen, J. I. and Bjorkhem, I. (1988). Biochem. J. 250, 521.
Henry, H. 1. and Norman, A. W.(1984). Ann. Rev. Nutr. 4,493. Hollis, B. W. and Napoli, J. L. (1985). Clin. Chem. 31, 1815. Holmberg, I . , Saarem, K., Pedersen, J. 1. and Bjorkhem, 1. (1986). Anal. Biochem. 159,317. Horst, R. L. (1985). In Vitamin D: Basic and Clinical Aspects, ed. by Kumar, R., p. 423. Martinus Nijhoff, Boston. Horst, R. L., Reinhardt, T. A., Beitr, D. C. and Littledike, E. T. (1981). Steroids 37, 581. Jongen, M. J. M., Van der Vijgh, W. J. F., Willems, H. J. J. and Netelenbos, J. C. (1981). Clin. Chem. 27, 444. Jongen, M. J. M., Van Ginkel, F. C., Van der Vijgh, W. J. F., Kuiper, S., Netelenbos, J. C. and Lips P. (1984). Clin. Chem. 30. .., 399. .._ Kim, H. Y., Yergey, J. A. and Salem, N. Jr. (1987). J. Chromafogr. 394,155. Mayer. E. and Schmidt-Gayk, H. (1984). Clin. Chem. 30,1199. Oftenbro, H., Falch, J. A., Holmberg, I. and Haug, E. (1988). Clin. Chim. Acta 176, 157. Okano, T., Mizuno, N., Shida. S., Takahashi, N., Kobayashi, T., Kuroda, E., Kodama, S. and Matsuo, T. (1981). J. Nutr. Sci. Vitamin 27, 43. Parker, C. E., de Wit, J. S. M., Smith, R. W., Gopinathan, M. B., Hernandez, 0..Tomer, K. 6..Vestal, C. H., Sanders, J. M . and Bend, J. R. (1988). Biomed. Environ. Mass Spectrom. 15, 623.
160
D. WATSON, R . ROSS AND K.D. R.SETCHELL
Paulson, S. K. and DeLoca, H. F. (1985). J. Biof. Chem. 260, 11488. Porteous, C. E., Coldwell, R. D., Trafford, D. J. H. and Makin, H. L. J. (1987). J. Steroid Biochern. 28, 785. Richmond, R., Clarke, S. R., Watson, D., Chappell, C. G.,Dollery, C. T. and Taylor, G. W. (1986). Biochim. Biophys. Acta 881, 159. Seamark, D.A., Trafford, D.J. H. and Makin, H. L.J.(1980). Clin, Chim. Acta 106, 51. Seamark, D. A., Trafford, D. J. H. and Makin, H. L. J. (1981). J. Steroid Biochem. 14, 111. Setchell, K. D. R. and Vestal, C. H. (1989). J. Lipid Res. Setchell, K. D. R., Welsh, M. 6. and Lim, C. K. (1987). J. Chromatogr. 386. 315.
Voyksner, R. D. and Bush, E. D.(1987). Biomed. Environ. Mass Spectrom. 14, 213. Watson, D., Taylor, G.W. and Murray, S. (1985). Biorned. Mass Spectrorn. 12, 610. Watson, D.. Taylor, G. W. and Murray, S. (1986). Biomed. Environ. Mass Spectrom. 13, 65. Watson, D., Taylor, G. W., Laird, S. and Vinson, G. P. (1987). Biochem. J. 242, 109. Watson, D., Taylor, G. W., Wilson, R., Cole, P. J. and Rowe, C. (1988). Biomed. Environ. Mass Spectrom. 17, 251. Watson, D., Ross R. and Setchell, K. D. R. (1990). Proceedings of the 38th ASMS Meeting on Mass Spectrometry and Allied Topics.
