ARCHIVES
OF BIOCHEMISTRY
Vol. 298, No. 2, November
AND
BIOPHYSICS
1, pp. 658666,
1992
Analysis of the Nucleoside Moiety of Cobalamin and Cobalamin Analogues Using Gas ChromatographyMass Spectrometry’ David
P. Sundin*
and Robert
H. Allen*,t,2
*Division of Hematology, Department of Medicine, and TDepartment of Biochemistry, University of Colorado Health Sciences Center, Denver, Colorado 80262
Received
May
1, 1992, and in revised
form
July
Academic
Press,
and Genetics,
15, 1992
Existing techniques for identification of cobalamin and cobalamin analogues generally use the intact molecule during characterization with somewhat ambiguous results. In this study a method is described for the identification of the nucleoside in the lower axial ligand of cobalamin and a variety of naturally occurring cobalamin analogues that differ from cobalamin in the base that is present in the nucleoside. Cobalamin and cobalamin analogues were isolated from biological samples by affinity chromatography using R-protein-Sepharose columns. The nucleosides of the lower axial ligand were then hydrolyzed and isolated by column chromatography using a mixed bed column. Nucleosides were oxidized with periodate and reduced with borohydride. After reisolation, the t-butyldimethylsilyl derivatives were prepared and analyzed using gas chromatography/ mass spectrometry with selected ion monitoring. A stable isotope internal standard of cobalamin was biosynthetically produced and used to quantitate cobalamin in rabbit kidney. Cobalamin analogues were also shown to be present in rabbit kidney, but they contain the 5,6dimethylbenzimidazole nucleoside (cY-ribazole) in the lower axial ligand, indicating that these analogues differ from cobalamin in thecorrin ring region of the molecule. 0 1992
Biophysics,
Inc.
i This work was supported by Department of Health and Human Services Research Grant DK21365, awarded by the National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health. ‘To whom correspondence should be addressed at Division of Hematology, University of Colorado Health Sciences Center, 4200 East Ninth Avenue, Campus Box B170, Denver, Colorado 80262. Fax: (303) 270-8477.
Cobalamin (Cb1,3vitamin ES& is an essential cofactor in the metabolism of mammals. Bacteria require Cbl as a coenzyme for many different reactions (l-3). Mammals, however, require Cbl as a coenzyme for only two enzymes, L-methylmalonyl CoA mutase and methionine synthase (4). The mutase utilizes the deoxyadenosyl form of Cbl and catalyzes the interconversion of L-methylmalonyl CoA to succinyl CoA. Methionine synthase uses the methyl form of Cbl and converts 5-methyltetrahydrofolate and homocysteine to tetrahydrofolate and methionine, respectively. The natural production of Cbl occurs exclusively in microorganisms. In addition, naturally occurring analogues of Cbl are common in nature as a result of microorganismal synthesis. The only difference between native Cbl and these analogues lies in the lower axial ligand or nucleotide loop portion of the molecule. Cbl contains 5,6dimethylbenzimidazole in the nucleotide of the lower axial ligand while a number of other bases are substituted in the naturally occurring Cbl analogues. Although these analogues can be more active for their systems in bacteria (5), the evidence suggests that some Cbl analogues may be inactive or even toxic in animals (6, 7). Recent studies have detected a number of other naturally occurring Cbl analogues of unknown origin in mammalian plasma and a variety of tissues (8-10). These newly recognized analogues were shown to be different from the more commonly known analogues of microorganismal origin. It has not been determined whether these new analogues differed from native Cbl in the same region as the analogues of microorganismal origin or at some other site. 3 Abbreviations used: Cbl, cobalamin; MTBSTFA, N-methyl-(tertbutyldimethylsilyl)trifluoroacetamide; TBDMS, tert-butyldimethylsilyl; BZA, benzimidazole; Cba, cobamide.
658 All
Copyright 0 1992 rights of reproduction
0003-9&x1/92 $5.00 by Academic Press, Inc. in any form reserved.
ANALYSIS
OF
COBALAMIN
In addition, the effect these newly recognized analogues have on Cbl metabolism remains unknown. In the past, studies involving Cbl and Cbl analogues have used the intact molecule for analysis (8, 9, 11, 12). The methods used for characterization involved difficult time-consuming chromatographic and electrophoretic procedures. Recently, high pressure liquid chromatography (HPLC) has been applied for the separation and characterization of Cbl and Cbl analogues (13-20). However, even with this technique, overlapping of retention times and marked similarities in absorption spectra have resulted in ambiguities in the identification of some analogues. NMR has also been used to analyze intact molecules of Cbl and Cbl analogues (21, 22). In this study, a methodology was developed to analyze the nucleoside of the lower axial ligand separate from the corrin ring portion of the Cbl and Cbl analogue molecules. The liberated nucleosides were isolated in a single step procedure. They were then oxidized with periodate, reduced with borohydride, and identified and quantitated using gas chromatography/mass spectroscopy (GC/MS). In addition, a stable isotope form of Cbl was biosynthetically produced and used to characterize and quantitate Cbl and Cbl analogues in rabbit kidney. MATERIALS
AND
METHODS
CM-Sephadex (C-25, 40-120 pm), D(+)-glucose, and yeast extract were obtained from Sigma Chemical Co. (St. Louis, MO). Octadecylsilane, 120-A pore size, 63-210 mesh (C,s), was purchased from YMC, Inc. (Morris Plains, NJ). Polypropylene 0.9 X 6.3-cm (4 ml) disposable columns with a 20-pm frit were obtained from Alltech North/Applied Science Labs (State College, PA). The bases: 5,6-[4,7-D,]dimethylbenzimidazole; 3,4-[2,5,6-D3]toluenediamine; and [2,3,4,5-D4]benzimidazole were custom synthesized and obtained from Merck, Sharpe, and Dohme Isotopes (St. Louis, MO). Propionibucterium shermanii (ATCC 9615) was obtained from the American Type Culture Collection. N-methyl(tert-Butyldimethylsilyl)trifluoroacetamide (MTBSTFA) was purchased from Pierce Chemical Co. (Rockford, IL). The hydrogen ion form of the CM-Sephadex was prepared by washing approximately 35 g with 1.2 liters of 0.17 N HCl, followed by a 500-ml H,O wash. The yield was approximately 200 ml of fully hydrated material. One percent HCN was prepared in a fume hood by passing 4 ml of a 300 mg/ml KCN solution over a 1.0 X 30-cm column filled with approximately 20 ml of Dowex 5OW-X8 (Bio-Rad Laboratories, Richmond, CA) in the hydrogen ion form. HCN was eluted by washing with H20. The first 12.5 ml of the wash was discarded. The next 50 ml was collected as 1% HCN. After collection, the HCN was kept at -20°C until used. Column preparation. (&s/CM-Sephadex columns were prepared for nucleoside isolation as follows. CIB, 0.5 cm in height, was measured into 0.9 X 6.3~cm disposable columns. An additional 20-pm frit was placed on top of the Cl8 to keep it in place. The Cla was activated by washing with 4 ml methanol, followed by a 4-ml H,O wash. A 0.6-ml aliquot of a 50/50 mix of CM-Sephadex in the hydrogen ion form/HP was pipetted on top of the Cl8 and washed with 4 ml H20. Cerous hydroxide hydrolysis. Hydrolysis of the nucleoside in the lower axial ligand was carried out using a modification (23) of the cerous hydroxide technique described by Friedrich and Bernhauer (24). Cbl or Cbl analogue (5 pg) was dried by vacuum centrifugation in a Savant (Farmingdale, NY) Speed-Vat in 1.5.ml lock-top Eppendorf microcentrifuge tubes. While drying, the cerous hydroxide (Ce[OH],) solution
AND
COBALAMIN
ANALOGUES
659
was prepared. Cerous nitrate, 64 mg, was weighed into a 1.5-ml centrifuge tube. NH,OH (750 ~1 of a 6.67% solution in H1O) was added, and the solution was mixed well and then centrifuged for 2 min in a Fisher (Pittsburgh, PA) Model 235A microcentrifuge. The supernatant was discarded and the pellet resuspended in 1 ml of 0.3% NH,OH followed by centrifugation as above. This washing step was repeated three more times. After the last wash, the pellet was brought up in 1 ml Hz0 and 20 ~1 of a 1% HCN solution was added (see above). Then 50 gl of the Ce[OH13 suspension was aliquotted onto the dried Cbl and Cbl analogue samples and they were incubated for 45 min at 90°C. Samples were vigorously mixed every 15 min. After cooling to room temperature, 1 ml of 2% acetic acid was added followed by mixing and centrifugation for 2 min in the microcentrifuge. Supernatants were then removed and applied to small individual C&M-Sephadex columns prepared as described above. Each column was washed with 4 ml of 0.1% HCN, followed by a 4-ml Hz0 wash. Nucleosides were eluted into 4-ml polypropylene conical tubes with two 0.55-ml aliquots of 4 N NH,OH in MeOH and dried by vacuum centrifugation. Oxidation/reduction. This step was carried out using modifications of techniques described by others (25,26). For oxidation of the hydrolyzed nucleosides, 50 bl of 8 mM sodium periodate in 0.015% NHIOH was added to the dried samples from above. They were incubated for 30 min at 25°C in the dark, with mixing every 10 min. Then, 50 ~1 of 40 mM sodium borohydride was added for reduction and incubation continued for 1 h at 25°C. Samples were mixed every 20 min and after 1 h, 1 ml of 2% acetic acid solution was added. The samples were mixed and spun in a microcentrifuge as above. The supernatants were applied to and eluted from the C&/CM-Sephadex columns and dried as described above. Deriuatization. The t-butyldimethylsilyl (TBDMS) derivatives of the oxidized/reduced nucleosides were prepared by adding 30 ~1 of a 1:2 mixture of MTBSTFA and acetonitrile to the dried samples, capping the vials, and incubating for 2 h at 90°C. Gas chromatography/mass spectrometry. Analysis of derivatized samples was performed on a Hewlett-Packard 5890 GC/5970 MS equipped with a 7673A autosampler system. A Supelco (Bellefonte, PA) SPB-1 (nonpolar, methyl-silicone) GC column was used for all experiments. The column was 10 m in length with an internal diameter of 0.25 mm and a film thickness of 0.25 pm. All experiments were carried out under standard autotune conditions with an injection port temperature of 300°C and a column head pressure of 7.5 psi. Initial temperature of the gas chromatograph oven was 80°C which was held for 1 min after sample injection. The temperature was then increased at a rate of 3O”C/ min until 300°C was reached where it was again held for 1 min. Data were collected from 6 to 10 min using a single ion monitor (SIM) program designed to monitor the ions with m/z equal to the molecular ions minus 57 (M-57) and minus 319 (M-319), which are specific for Chl and each Cbl analogue (see Results). R-protein-Sepharose. Hog R-protein-Sepharose was prepared and utilized for the purification of Cbl and Cbl analogues as described previously (27), except that hog intrinsic factor was not removed. The final material had a Cbl binding capacity of 250 pg of Cbl per milliliter of Sepharose and contained 85% R-protein and 15% intrinsic factor. Bacterial production of Cbl and Cbl analagues. Naturally occurring Cbl analogues with nucleoside moieties different from Cbl were produced by bacterial fermentation using the method of Perlman and Barrett (28) with minor modifications described previously (27). Bases were added to the culture medium in dry form prior to autoclaving and CoCl, was used rather than Co(NO&. The following respective bases were used to obtain the following Cbl analogues: henzimidazole (BZA), [BZAICNcobamide (Cba); 3,4-toluenediamine, [5(6)-MeBZA]CN-Cba (the notation . . 5(6)-Me . . refers to the fact that the methyl group may be in either the 5 or the 6 position); 5(6)methoxyBZA, [5(6)-methoxyBZA]CN-Cba; 5(6)-hydroxyBZA, [5(6)-OHBZAICN-Cba; adenine, [Ade]CN-Cba; and 2-methyladenine, [2-MeAde]CN-Cba. The deuterated bases [2,3,4,5-D,]benzimidazole, 3,4-[2,5,6-D,]tolu-
660
SUNDIN
AND
enediamine, and 5,6-[4,7-D,]dimethylbenzimidazole were used to obtain [[4,5,6,7-D,]BZA]CN-Cba, [5(6)-[4,5(6),7-D,]MeBZA]CN-Cba, and [[4,7-D&,6-diMeBZA]CN-Cba (i.e., [D,]CN-Cbl), respectively. Cbl-like material was purified from bacterial lysates using hog R-proteinSepharose as described above. Fecal sample preparation. Animal fecal samples (15 g) were added to a Waring Blender and homogenized until uniform in 50 ml of 0.01 M KHzPOl containing 0.1 mg/ml KCN. The solution was then boiled for 30 min with mixing every 10 min. Samples were allowed to cool and then spun for 30 min at 14,000 rpm (30,OOOg) in a Beckman J-21C centrifuge. The supernatants were applied to columns containing 5 ml of C,,. Columns were washed with 25 ml of Hz0 and eluted with 5 ml of methanol. Elutions were dried overnight at 60°C by vacuum centrifugation. The residue was brought up in 1 ml of Hz0 and applied to 0.5 ml R-protein-Sepharose columns. The columns were washed with 2 ml of a 0.1 M Glycine-NaOH (pH 10)/l M NaCl solution and 2.5 ml of HzO. Cbl-like material was then eluted with 1 ml of liquified phenol. Samples were diluted 1:4 with l-butanol and dried overnight at 60°C by vacuum centrifugation. Dried residues were brought up in 1 ml of Hz0 and the concentration of Cbl and Cbl analogues was estimated spectrophotometrically on a Hewlett-Packard 8425A DA spectrophotometer/ computer using the absorption at 360 and 550 nm and extinction coefficients of 28,100 and 8700, respectively (29). Preparation and analysis of Cbl-like material from rabbit kidney. Frozen, mature rabbit kidney (300 g, Pelfreeze) was homogenized with a Waring commercial blender for 3 min at 4°C in 900 ml of 0.17 M sodium acetate-HCl, pH 4.5, containing 0.10 M NaCl and 45 mg of KCN. Just prior to homogenization, 100 pg of [D2]CN-Cbl and 1 &i of CN-[57Co]Cbl (10 aC!i per 70 ng of CN-Cbl, Amersham, Chicago Heights, IL) were added to the solution to make it possible to quantitate the amount and recovery of endogenous Cbl, respectively. The homogenate was heated for 45 min at 85-90°C and then cooled to 4°C. It was then spun for 30 min at 30,OOOg. The supernatant was removed and an aliquot counted in a Beckman gamma 4000 counter for determination of recovery, which was 93% at this point. Isolation of Cbl-like material from kidney homogenate was performed as described (27) using 6 ml of R-protein-Sepharose with the following modifications. Cbl-like material was eluted with 12 ml of 60% pyridine and then treated with 1 ml of 1% HCN (see above). The solution was incubated for at least 30 min at 25°C in the dark, then dried by vacuum centrifugation. The dried Chl-like material was counted for estimation of recovery as above and then brought up in Hz0 and the concentration was estimated spectrophotometrically as described above. Recovery of starting material was estimated to be 73% and the concentration was approximately 350 ag/ml in a total volume of 1 ml. Once this analysis was completed, the Cbl-like material was redried. The redried Cbl-like material was then brought up in 250 ~1 of HzO, 10% was removed for analysis, and the rest was analyzed by HPLC using a Hewlett-Packard 1090 HPLC/9153 computer system to monitor absorbance at 360 nm. All of the sample, 225 ~1, was injected into a Keystone (Keystone Scientific Inc., Bellefonte, PA) 2 X 250-mm PRP1 lo-Frn particle size column equilibrated with 70% solution A (1% acetic acid in H,O) and 30% solution B (20% isopropanol and 1% acetic acid in H,O). Elution was begun immediately using a linear gradient that started at the original 70/30 mix and continued for 35 min, at which point the solvent mix was 60% solution A and 40% solution B. A second linear gradient was then begun such that 100% solution B was reached at 45 min where it was maintained for an additional 5 min. The flow rate was 0.3 ml/min and, once elution was begun, 0.45-ml fractions were collected. OH-Cbl elutes in the void volume, i.e., fractions 2-3, in this system. Each fraction was dried by vacuum centrifugation, brought up in 1 ml H,O, and assayed for radioactivity and absorbance at 360 nm. Based on radioactivity, the recovery was 59%. Approximately 5 pg of Cbl-like material from each fraction was aliquotted into 1.5-ml locktop microcentrifuge tubes which were dried, hydrolyzed, oxidized, reduced, derivatized, and analyzed by GC/MS as described above.
ALLEN In the experiment employing rabbit kidney, the amount of endogenous Cbl was calculated based on the ratio of the abundance of ions M-319 and M-57 for endogenous CN-Cbl to the abundance of the corresponding ions for [Da]CN-Cbl, which was added to the rabbit kidney in a known amount. In calculating the ratio of the M-319 ion for CN-Cbl (m/z = 303) to that of the M-319 ion for [D2]CN-Cbl (m/z = 305), the abundance of the 305 ion was corrected for the amount contributed to it by endogenous CN-Cbl as a result of naturally occurring isotope abundance. This correction, which was obtained from data obtained with unenriched CN-Cbl, represented 7.1% of the m/z 303 peak being present as a 305 peak. A similar correction was employed for the ratio of the M-57 ion for CN-Cbl (m/z = 565) to that of the M-57 ion for [D,]CN-Cbl (m/z = 567). Here 21.0% of the m/z 565 peak was present as a 567 peak. In addition, a further correction was required to obtain the appropriate abundance for the M-319 ion for CN-Cbl (m/z = 303) since an amount equal to 6.5% of the corrected abundance for the M-319 ion for [D,]CNCbl (m/z = 305) was present as a m/z = 303 fragment that did not contain the benzene ring portion of the nucleoside (see Results). Thus, the calculated abundances used for quantitating CN-Cbl in rabbit kidney were obtained as follows: (corrected
abundance
305) = (observed
(corrected
abundance
567) = (observed
abundance
305)
- 0.071 X (observed abundance
abundance
303)
abundance
565)
abundance
305).
567)
~ 0.210 X (observed (corrected
abundance
303)
= (observed - 0.065
abundance X (corrected
303)
As an example, if the value for the ratio (corrected abundance 303)/ (corrected abundance 305) were 1.00, this would indicate that the amount of endogenous Cbl and Cbl analogue containing the 5,6-dimethylbenzimidazole nucleoside was the same as the amount of the [D,]CN-Cbl added to the sample. A value of 10.0 for the ratio would indicate that the endogenous amount was 10 times greater than the added amount of [Ds]CN-Cbl.
RESULTS
Figure 1 shows the structure of CN-Cbl and the partial structures of six naturally occurring bacterial Cbl analogues. Only the portion of the lower axial ligand that differs from native Cbl is shown. In Fig. 2 can be seen the general scheme we used to liberate the nucleoside portion of the lower axial ligand/nucleotide loop. This technique cleaves the phosphodiester bond on either side of the phosphate group and disrupts the cobalt/nitrogen coordination producing cobinamide, inorganic phosphate, and the nucleoside. As illustrated in Fig. 1, basesof the naturally occurring analogues are either uncharged at neutral pH, as for the base in native Cbl, or partially negatively or partially positively charged. Because of this, mixed bed columns were used after the cerous hydroxide hydrolysis and the oxidation/reduction procedures. Samples were applied to the mixed bed columns at low pH (2% acetic acid) to ensure that the hydroxyl group of the nucleoside from [5(6)OHBZAICN-Cba would not be ionized and therefore be retained by the C1sportion of the column, and that the
ANALYSIS
OF
COBALAMIN
AND
COBALAMIN
661
ANALOGUES CN
Lower CN-Cbl Alternative
Bases
Axial
Ligand
45’,90°
of Naturally
Occurring
of CN-Cbl
1Ce(Ol-03
Analogues
t
Uncharged
eN” OHM0I
0
\ /O*TP\ 0 0 ti ttOCH* [BZA]
CN-Cba
[S,(6)-MeBZA]
CNmCba
[En.@-OCH,BZA]
CN-Cba
,norganic
Cobmamide Partially
Phosphate
Charged
0
5.6~Dnrethylbenzimldazole Nucleoside
FIG. 2. Scheme for cerous hydroxide hydrolysis of the nucleoside in the lower axial ligand of CN-Cbl. It is the same for Cbl analogues except that different bases are present in the nucleoside. Only the D ring and lower axial ligand of the intact molecule are shown. [s.(6)-OHBZA]
CN-Cba
[Ade]
CN-Cba
[P-MeAde]
FIG. 1. Structure of CN-CM (top) and partial structures urally occurring analogues (bottom). Only the portions of the that differ from native Cbl are shown (i.e., the base of the in the lower axial ligand). The top three bases are uncharged pH, while the bottom three are partially negatively charged of 5,(6)-OHBZA, and partially positively charged in the case adenine containing bases.
CN-Cba
of six natanalogues nucleoside at neutral in the case of the two
amino groups of the nucleosides from [Ade]CN-Cba and [2-MeAde]CN-Cba would be fully protonated and therefore be retained by the cationic CM-Sephadex portion of the column. In preliminary experiments, we found that the furanose ring structure of the ribose moiety was very resistant to derivatization with MTBSTFA, even at high temperatures and extended lengths of time. We felt the explanation for this resistance was steric hindrance resulting from hydroxyl groups on the adjacent 2’ and 3’ carbon atoms of the ribose moiety (see Fig. 2). We reasoned that opening up the ring structure would help to overcome the resistance problem. To achieve this, we cleaved the furanose ring between the 2’ and 3’ carbon atoms using a periodate oxidation/borohydride reduction scheme. Once oxidized and reduced, the modified nucleoside was reisolated, readily derivatized, and analyzed by GC/MS. Additional support for the schemata illustrated in Fig. 2 was obtained from experiments that showed that the M-57 and M-319 ions for Cbl and Cbl analogues were not obtained when either the periodate or the borohydride steps were omitted or when their order was reversed.
Spectra for the nucleosides obtained from unlabeled CN-Cbl and [D2]CN-Cbl containing 5,6-[4,7-Dz]dimethylbenzimidazole are shown in Fig. 3. In the case of the unlabeled nucleoside, major ions were observed with m/z of 245,287,303,393,477, and 565. Every one of these ions, except for 287, appears to contain at least a portion of the benzene ring since the values for m/z were increased by a value of 2 when the labeled nucleoside was analyzed. An ion with m/z 303 was also present in the spectrum
A
B
303 T
100
so 2 E c
so
I m .z I
40
z
20
0
FIG. 3. Spectra of the nucleosides (B) after they were oxidized, reduced, Materials and Methods.
from CN-Cbl and derivatized
(A) and D,-CN-Cbl as described under
662
SUNDIN
from the labeled nucleoside and represented 6.5% of the abundance of the 305 ion. This ion arises from a portion of the nucleoside that does not contain the benzene portion of the base rather than from contamination with unlabeled nucleoside because detectable (~1%) amounts of ions 245, 393,477, and 565 were not present in the spectrum of the labeled nucleoside. Figure 4 shows the proposed structure and fragmentation points of the two most abundant ions for the fully derivatized 5,6-dimethylbenzimidazole nucleoside of native Cbl. The larger 565 ion represents the fully derivatized, M-57 form of the nucleoside. This form represents a species in which a t-butyl fragment (atomic mass = 57) has been lost at any one of the positions x, y, or z. The smaller 303 ion represents a form of the nucleoside in which the ribose ring has been cleaved, with a fragment of the ribose ring remaining attached to the base portion of the nucleoside. It was this ribose fragment/base form (M-319) that was consistent and was monitored with the M-57 form. The M-57 and M-319 ions proved to be descriptive of each nucleoside, the only difference in atomic mass being attributed to variation in the base of the nucleoside. Figure 5 shows chromatograms of an experiment in which the nucleosides from CN-Cbl (top), [5(6)-OHBZAICN-Cba (middle), and the nucleoside fl-adenosine (bottom), the common form of adenosine, were subjected to hydrolysis conditions, oxidized with periodate, reduced with borohydride, derivatized, and analyzed by GC/MS. These three nucleosides represent all three ionic forms of nucleosides which result from Ce(OH)3 hydrolysis. As can be seen in Fig. 5, each nucleoside being analyzed eluted from the gas chromatography column at times that were clearly distinguishable from each other. We also applied our technique to a mixture of intact CN-Cbl and six naturally occurring bacterial Cbl analogues. Table I shows the predicted ions monitored and retention times for all seven nucleosides. It was clear from
(Y)
AND
ALLEN
1
6.0
7.0 Ion 303.20
60 Ion 667.40 G r E a
OF BIOCHEMISTRY
Vol. 298, No. 2, November
AND
BIOPHYSICS
1, pp. 658666,
1992
Analysis of the Nucleoside Moiety of Cobalamin and Cobalamin Analogues Using Gas ChromatographyMass Spectrometry’ David
P. Sundin*
and Robert
H. Allen*,t,2
*Division of Hematology, Department of Medicine, and TDepartment of Biochemistry, University of Colorado Health Sciences Center, Denver, Colorado 80262
Received
May
1, 1992, and in revised
form
July
Academic
Press,
and Genetics,
15, 1992
Existing techniques for identification of cobalamin and cobalamin analogues generally use the intact molecule during characterization with somewhat ambiguous results. In this study a method is described for the identification of the nucleoside in the lower axial ligand of cobalamin and a variety of naturally occurring cobalamin analogues that differ from cobalamin in the base that is present in the nucleoside. Cobalamin and cobalamin analogues were isolated from biological samples by affinity chromatography using R-protein-Sepharose columns. The nucleosides of the lower axial ligand were then hydrolyzed and isolated by column chromatography using a mixed bed column. Nucleosides were oxidized with periodate and reduced with borohydride. After reisolation, the t-butyldimethylsilyl derivatives were prepared and analyzed using gas chromatography/ mass spectrometry with selected ion monitoring. A stable isotope internal standard of cobalamin was biosynthetically produced and used to quantitate cobalamin in rabbit kidney. Cobalamin analogues were also shown to be present in rabbit kidney, but they contain the 5,6dimethylbenzimidazole nucleoside (cY-ribazole) in the lower axial ligand, indicating that these analogues differ from cobalamin in thecorrin ring region of the molecule. 0 1992
Biophysics,
Inc.
i This work was supported by Department of Health and Human Services Research Grant DK21365, awarded by the National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health. ‘To whom correspondence should be addressed at Division of Hematology, University of Colorado Health Sciences Center, 4200 East Ninth Avenue, Campus Box B170, Denver, Colorado 80262. Fax: (303) 270-8477.
Cobalamin (Cb1,3vitamin ES& is an essential cofactor in the metabolism of mammals. Bacteria require Cbl as a coenzyme for many different reactions (l-3). Mammals, however, require Cbl as a coenzyme for only two enzymes, L-methylmalonyl CoA mutase and methionine synthase (4). The mutase utilizes the deoxyadenosyl form of Cbl and catalyzes the interconversion of L-methylmalonyl CoA to succinyl CoA. Methionine synthase uses the methyl form of Cbl and converts 5-methyltetrahydrofolate and homocysteine to tetrahydrofolate and methionine, respectively. The natural production of Cbl occurs exclusively in microorganisms. In addition, naturally occurring analogues of Cbl are common in nature as a result of microorganismal synthesis. The only difference between native Cbl and these analogues lies in the lower axial ligand or nucleotide loop portion of the molecule. Cbl contains 5,6dimethylbenzimidazole in the nucleotide of the lower axial ligand while a number of other bases are substituted in the naturally occurring Cbl analogues. Although these analogues can be more active for their systems in bacteria (5), the evidence suggests that some Cbl analogues may be inactive or even toxic in animals (6, 7). Recent studies have detected a number of other naturally occurring Cbl analogues of unknown origin in mammalian plasma and a variety of tissues (8-10). These newly recognized analogues were shown to be different from the more commonly known analogues of microorganismal origin. It has not been determined whether these new analogues differed from native Cbl in the same region as the analogues of microorganismal origin or at some other site. 3 Abbreviations used: Cbl, cobalamin; MTBSTFA, N-methyl-(tertbutyldimethylsilyl)trifluoroacetamide; TBDMS, tert-butyldimethylsilyl; BZA, benzimidazole; Cba, cobamide.
658 All
Copyright 0 1992 rights of reproduction
0003-9&x1/92 $5.00 by Academic Press, Inc. in any form reserved.
ANALYSIS
OF
COBALAMIN
In addition, the effect these newly recognized analogues have on Cbl metabolism remains unknown. In the past, studies involving Cbl and Cbl analogues have used the intact molecule for analysis (8, 9, 11, 12). The methods used for characterization involved difficult time-consuming chromatographic and electrophoretic procedures. Recently, high pressure liquid chromatography (HPLC) has been applied for the separation and characterization of Cbl and Cbl analogues (13-20). However, even with this technique, overlapping of retention times and marked similarities in absorption spectra have resulted in ambiguities in the identification of some analogues. NMR has also been used to analyze intact molecules of Cbl and Cbl analogues (21, 22). In this study, a methodology was developed to analyze the nucleoside of the lower axial ligand separate from the corrin ring portion of the Cbl and Cbl analogue molecules. The liberated nucleosides were isolated in a single step procedure. They were then oxidized with periodate, reduced with borohydride, and identified and quantitated using gas chromatography/mass spectroscopy (GC/MS). In addition, a stable isotope form of Cbl was biosynthetically produced and used to characterize and quantitate Cbl and Cbl analogues in rabbit kidney. MATERIALS
AND
METHODS
CM-Sephadex (C-25, 40-120 pm), D(+)-glucose, and yeast extract were obtained from Sigma Chemical Co. (St. Louis, MO). Octadecylsilane, 120-A pore size, 63-210 mesh (C,s), was purchased from YMC, Inc. (Morris Plains, NJ). Polypropylene 0.9 X 6.3-cm (4 ml) disposable columns with a 20-pm frit were obtained from Alltech North/Applied Science Labs (State College, PA). The bases: 5,6-[4,7-D,]dimethylbenzimidazole; 3,4-[2,5,6-D3]toluenediamine; and [2,3,4,5-D4]benzimidazole were custom synthesized and obtained from Merck, Sharpe, and Dohme Isotopes (St. Louis, MO). Propionibucterium shermanii (ATCC 9615) was obtained from the American Type Culture Collection. N-methyl(tert-Butyldimethylsilyl)trifluoroacetamide (MTBSTFA) was purchased from Pierce Chemical Co. (Rockford, IL). The hydrogen ion form of the CM-Sephadex was prepared by washing approximately 35 g with 1.2 liters of 0.17 N HCl, followed by a 500-ml H,O wash. The yield was approximately 200 ml of fully hydrated material. One percent HCN was prepared in a fume hood by passing 4 ml of a 300 mg/ml KCN solution over a 1.0 X 30-cm column filled with approximately 20 ml of Dowex 5OW-X8 (Bio-Rad Laboratories, Richmond, CA) in the hydrogen ion form. HCN was eluted by washing with H20. The first 12.5 ml of the wash was discarded. The next 50 ml was collected as 1% HCN. After collection, the HCN was kept at -20°C until used. Column preparation. (&s/CM-Sephadex columns were prepared for nucleoside isolation as follows. CIB, 0.5 cm in height, was measured into 0.9 X 6.3~cm disposable columns. An additional 20-pm frit was placed on top of the Cl8 to keep it in place. The Cla was activated by washing with 4 ml methanol, followed by a 4-ml H,O wash. A 0.6-ml aliquot of a 50/50 mix of CM-Sephadex in the hydrogen ion form/HP was pipetted on top of the Cl8 and washed with 4 ml H20. Cerous hydroxide hydrolysis. Hydrolysis of the nucleoside in the lower axial ligand was carried out using a modification (23) of the cerous hydroxide technique described by Friedrich and Bernhauer (24). Cbl or Cbl analogue (5 pg) was dried by vacuum centrifugation in a Savant (Farmingdale, NY) Speed-Vat in 1.5.ml lock-top Eppendorf microcentrifuge tubes. While drying, the cerous hydroxide (Ce[OH],) solution
AND
COBALAMIN
ANALOGUES
659
was prepared. Cerous nitrate, 64 mg, was weighed into a 1.5-ml centrifuge tube. NH,OH (750 ~1 of a 6.67% solution in H1O) was added, and the solution was mixed well and then centrifuged for 2 min in a Fisher (Pittsburgh, PA) Model 235A microcentrifuge. The supernatant was discarded and the pellet resuspended in 1 ml of 0.3% NH,OH followed by centrifugation as above. This washing step was repeated three more times. After the last wash, the pellet was brought up in 1 ml Hz0 and 20 ~1 of a 1% HCN solution was added (see above). Then 50 gl of the Ce[OH13 suspension was aliquotted onto the dried Cbl and Cbl analogue samples and they were incubated for 45 min at 90°C. Samples were vigorously mixed every 15 min. After cooling to room temperature, 1 ml of 2% acetic acid was added followed by mixing and centrifugation for 2 min in the microcentrifuge. Supernatants were then removed and applied to small individual C&M-Sephadex columns prepared as described above. Each column was washed with 4 ml of 0.1% HCN, followed by a 4-ml Hz0 wash. Nucleosides were eluted into 4-ml polypropylene conical tubes with two 0.55-ml aliquots of 4 N NH,OH in MeOH and dried by vacuum centrifugation. Oxidation/reduction. This step was carried out using modifications of techniques described by others (25,26). For oxidation of the hydrolyzed nucleosides, 50 bl of 8 mM sodium periodate in 0.015% NHIOH was added to the dried samples from above. They were incubated for 30 min at 25°C in the dark, with mixing every 10 min. Then, 50 ~1 of 40 mM sodium borohydride was added for reduction and incubation continued for 1 h at 25°C. Samples were mixed every 20 min and after 1 h, 1 ml of 2% acetic acid solution was added. The samples were mixed and spun in a microcentrifuge as above. The supernatants were applied to and eluted from the C&/CM-Sephadex columns and dried as described above. Deriuatization. The t-butyldimethylsilyl (TBDMS) derivatives of the oxidized/reduced nucleosides were prepared by adding 30 ~1 of a 1:2 mixture of MTBSTFA and acetonitrile to the dried samples, capping the vials, and incubating for 2 h at 90°C. Gas chromatography/mass spectrometry. Analysis of derivatized samples was performed on a Hewlett-Packard 5890 GC/5970 MS equipped with a 7673A autosampler system. A Supelco (Bellefonte, PA) SPB-1 (nonpolar, methyl-silicone) GC column was used for all experiments. The column was 10 m in length with an internal diameter of 0.25 mm and a film thickness of 0.25 pm. All experiments were carried out under standard autotune conditions with an injection port temperature of 300°C and a column head pressure of 7.5 psi. Initial temperature of the gas chromatograph oven was 80°C which was held for 1 min after sample injection. The temperature was then increased at a rate of 3O”C/ min until 300°C was reached where it was again held for 1 min. Data were collected from 6 to 10 min using a single ion monitor (SIM) program designed to monitor the ions with m/z equal to the molecular ions minus 57 (M-57) and minus 319 (M-319), which are specific for Chl and each Cbl analogue (see Results). R-protein-Sepharose. Hog R-protein-Sepharose was prepared and utilized for the purification of Cbl and Cbl analogues as described previously (27), except that hog intrinsic factor was not removed. The final material had a Cbl binding capacity of 250 pg of Cbl per milliliter of Sepharose and contained 85% R-protein and 15% intrinsic factor. Bacterial production of Cbl and Cbl analagues. Naturally occurring Cbl analogues with nucleoside moieties different from Cbl were produced by bacterial fermentation using the method of Perlman and Barrett (28) with minor modifications described previously (27). Bases were added to the culture medium in dry form prior to autoclaving and CoCl, was used rather than Co(NO&. The following respective bases were used to obtain the following Cbl analogues: henzimidazole (BZA), [BZAICNcobamide (Cba); 3,4-toluenediamine, [5(6)-MeBZA]CN-Cba (the notation . . 5(6)-Me . . refers to the fact that the methyl group may be in either the 5 or the 6 position); 5(6)methoxyBZA, [5(6)-methoxyBZA]CN-Cba; 5(6)-hydroxyBZA, [5(6)-OHBZAICN-Cba; adenine, [Ade]CN-Cba; and 2-methyladenine, [2-MeAde]CN-Cba. The deuterated bases [2,3,4,5-D,]benzimidazole, 3,4-[2,5,6-D,]tolu-
660
SUNDIN
AND
enediamine, and 5,6-[4,7-D,]dimethylbenzimidazole were used to obtain [[4,5,6,7-D,]BZA]CN-Cba, [5(6)-[4,5(6),7-D,]MeBZA]CN-Cba, and [[4,7-D&,6-diMeBZA]CN-Cba (i.e., [D,]CN-Cbl), respectively. Cbl-like material was purified from bacterial lysates using hog R-proteinSepharose as described above. Fecal sample preparation. Animal fecal samples (15 g) were added to a Waring Blender and homogenized until uniform in 50 ml of 0.01 M KHzPOl containing 0.1 mg/ml KCN. The solution was then boiled for 30 min with mixing every 10 min. Samples were allowed to cool and then spun for 30 min at 14,000 rpm (30,OOOg) in a Beckman J-21C centrifuge. The supernatants were applied to columns containing 5 ml of C,,. Columns were washed with 25 ml of Hz0 and eluted with 5 ml of methanol. Elutions were dried overnight at 60°C by vacuum centrifugation. The residue was brought up in 1 ml of Hz0 and applied to 0.5 ml R-protein-Sepharose columns. The columns were washed with 2 ml of a 0.1 M Glycine-NaOH (pH 10)/l M NaCl solution and 2.5 ml of HzO. Cbl-like material was then eluted with 1 ml of liquified phenol. Samples were diluted 1:4 with l-butanol and dried overnight at 60°C by vacuum centrifugation. Dried residues were brought up in 1 ml of Hz0 and the concentration of Cbl and Cbl analogues was estimated spectrophotometrically on a Hewlett-Packard 8425A DA spectrophotometer/ computer using the absorption at 360 and 550 nm and extinction coefficients of 28,100 and 8700, respectively (29). Preparation and analysis of Cbl-like material from rabbit kidney. Frozen, mature rabbit kidney (300 g, Pelfreeze) was homogenized with a Waring commercial blender for 3 min at 4°C in 900 ml of 0.17 M sodium acetate-HCl, pH 4.5, containing 0.10 M NaCl and 45 mg of KCN. Just prior to homogenization, 100 pg of [D2]CN-Cbl and 1 &i of CN-[57Co]Cbl (10 aC!i per 70 ng of CN-Cbl, Amersham, Chicago Heights, IL) were added to the solution to make it possible to quantitate the amount and recovery of endogenous Cbl, respectively. The homogenate was heated for 45 min at 85-90°C and then cooled to 4°C. It was then spun for 30 min at 30,OOOg. The supernatant was removed and an aliquot counted in a Beckman gamma 4000 counter for determination of recovery, which was 93% at this point. Isolation of Cbl-like material from kidney homogenate was performed as described (27) using 6 ml of R-protein-Sepharose with the following modifications. Cbl-like material was eluted with 12 ml of 60% pyridine and then treated with 1 ml of 1% HCN (see above). The solution was incubated for at least 30 min at 25°C in the dark, then dried by vacuum centrifugation. The dried Chl-like material was counted for estimation of recovery as above and then brought up in Hz0 and the concentration was estimated spectrophotometrically as described above. Recovery of starting material was estimated to be 73% and the concentration was approximately 350 ag/ml in a total volume of 1 ml. Once this analysis was completed, the Cbl-like material was redried. The redried Cbl-like material was then brought up in 250 ~1 of HzO, 10% was removed for analysis, and the rest was analyzed by HPLC using a Hewlett-Packard 1090 HPLC/9153 computer system to monitor absorbance at 360 nm. All of the sample, 225 ~1, was injected into a Keystone (Keystone Scientific Inc., Bellefonte, PA) 2 X 250-mm PRP1 lo-Frn particle size column equilibrated with 70% solution A (1% acetic acid in H,O) and 30% solution B (20% isopropanol and 1% acetic acid in H,O). Elution was begun immediately using a linear gradient that started at the original 70/30 mix and continued for 35 min, at which point the solvent mix was 60% solution A and 40% solution B. A second linear gradient was then begun such that 100% solution B was reached at 45 min where it was maintained for an additional 5 min. The flow rate was 0.3 ml/min and, once elution was begun, 0.45-ml fractions were collected. OH-Cbl elutes in the void volume, i.e., fractions 2-3, in this system. Each fraction was dried by vacuum centrifugation, brought up in 1 ml H,O, and assayed for radioactivity and absorbance at 360 nm. Based on radioactivity, the recovery was 59%. Approximately 5 pg of Cbl-like material from each fraction was aliquotted into 1.5-ml locktop microcentrifuge tubes which were dried, hydrolyzed, oxidized, reduced, derivatized, and analyzed by GC/MS as described above.
ALLEN In the experiment employing rabbit kidney, the amount of endogenous Cbl was calculated based on the ratio of the abundance of ions M-319 and M-57 for endogenous CN-Cbl to the abundance of the corresponding ions for [Da]CN-Cbl, which was added to the rabbit kidney in a known amount. In calculating the ratio of the M-319 ion for CN-Cbl (m/z = 303) to that of the M-319 ion for [D2]CN-Cbl (m/z = 305), the abundance of the 305 ion was corrected for the amount contributed to it by endogenous CN-Cbl as a result of naturally occurring isotope abundance. This correction, which was obtained from data obtained with unenriched CN-Cbl, represented 7.1% of the m/z 303 peak being present as a 305 peak. A similar correction was employed for the ratio of the M-57 ion for CN-Cbl (m/z = 565) to that of the M-57 ion for [D,]CN-Cbl (m/z = 567). Here 21.0% of the m/z 565 peak was present as a 567 peak. In addition, a further correction was required to obtain the appropriate abundance for the M-319 ion for CN-Cbl (m/z = 303) since an amount equal to 6.5% of the corrected abundance for the M-319 ion for [D,]CNCbl (m/z = 305) was present as a m/z = 303 fragment that did not contain the benzene ring portion of the nucleoside (see Results). Thus, the calculated abundances used for quantitating CN-Cbl in rabbit kidney were obtained as follows: (corrected
abundance
305) = (observed
(corrected
abundance
567) = (observed
abundance
305)
- 0.071 X (observed abundance
abundance
303)
abundance
565)
abundance
305).
567)
~ 0.210 X (observed (corrected
abundance
303)
= (observed - 0.065
abundance X (corrected
303)
As an example, if the value for the ratio (corrected abundance 303)/ (corrected abundance 305) were 1.00, this would indicate that the amount of endogenous Cbl and Cbl analogue containing the 5,6-dimethylbenzimidazole nucleoside was the same as the amount of the [D,]CN-Cbl added to the sample. A value of 10.0 for the ratio would indicate that the endogenous amount was 10 times greater than the added amount of [Ds]CN-Cbl.
RESULTS
Figure 1 shows the structure of CN-Cbl and the partial structures of six naturally occurring bacterial Cbl analogues. Only the portion of the lower axial ligand that differs from native Cbl is shown. In Fig. 2 can be seen the general scheme we used to liberate the nucleoside portion of the lower axial ligand/nucleotide loop. This technique cleaves the phosphodiester bond on either side of the phosphate group and disrupts the cobalt/nitrogen coordination producing cobinamide, inorganic phosphate, and the nucleoside. As illustrated in Fig. 1, basesof the naturally occurring analogues are either uncharged at neutral pH, as for the base in native Cbl, or partially negatively or partially positively charged. Because of this, mixed bed columns were used after the cerous hydroxide hydrolysis and the oxidation/reduction procedures. Samples were applied to the mixed bed columns at low pH (2% acetic acid) to ensure that the hydroxyl group of the nucleoside from [5(6)OHBZAICN-Cba would not be ionized and therefore be retained by the C1sportion of the column, and that the
ANALYSIS
OF
COBALAMIN
AND
COBALAMIN
661
ANALOGUES CN
Lower CN-Cbl Alternative
Bases
Axial
Ligand
45’,90°
of Naturally
Occurring
of CN-Cbl
1Ce(Ol-03
Analogues
t
Uncharged
eN” OHM0I
0
\ /O*TP\ 0 0 ti ttOCH* [BZA]
CN-Cba
[S,(6)-MeBZA]
CNmCba
[En.@-OCH,BZA]
CN-Cba
,norganic
Cobmamide Partially
Phosphate
Charged
0
5.6~Dnrethylbenzimldazole Nucleoside
FIG. 2. Scheme for cerous hydroxide hydrolysis of the nucleoside in the lower axial ligand of CN-Cbl. It is the same for Cbl analogues except that different bases are present in the nucleoside. Only the D ring and lower axial ligand of the intact molecule are shown. [s.(6)-OHBZA]
CN-Cba
[Ade]
CN-Cba
[P-MeAde]
FIG. 1. Structure of CN-CM (top) and partial structures urally occurring analogues (bottom). Only the portions of the that differ from native Cbl are shown (i.e., the base of the in the lower axial ligand). The top three bases are uncharged pH, while the bottom three are partially negatively charged of 5,(6)-OHBZA, and partially positively charged in the case adenine containing bases.
CN-Cba
of six natanalogues nucleoside at neutral in the case of the two
amino groups of the nucleosides from [Ade]CN-Cba and [2-MeAde]CN-Cba would be fully protonated and therefore be retained by the cationic CM-Sephadex portion of the column. In preliminary experiments, we found that the furanose ring structure of the ribose moiety was very resistant to derivatization with MTBSTFA, even at high temperatures and extended lengths of time. We felt the explanation for this resistance was steric hindrance resulting from hydroxyl groups on the adjacent 2’ and 3’ carbon atoms of the ribose moiety (see Fig. 2). We reasoned that opening up the ring structure would help to overcome the resistance problem. To achieve this, we cleaved the furanose ring between the 2’ and 3’ carbon atoms using a periodate oxidation/borohydride reduction scheme. Once oxidized and reduced, the modified nucleoside was reisolated, readily derivatized, and analyzed by GC/MS. Additional support for the schemata illustrated in Fig. 2 was obtained from experiments that showed that the M-57 and M-319 ions for Cbl and Cbl analogues were not obtained when either the periodate or the borohydride steps were omitted or when their order was reversed.
Spectra for the nucleosides obtained from unlabeled CN-Cbl and [D2]CN-Cbl containing 5,6-[4,7-Dz]dimethylbenzimidazole are shown in Fig. 3. In the case of the unlabeled nucleoside, major ions were observed with m/z of 245,287,303,393,477, and 565. Every one of these ions, except for 287, appears to contain at least a portion of the benzene ring since the values for m/z were increased by a value of 2 when the labeled nucleoside was analyzed. An ion with m/z 303 was also present in the spectrum
A
B
303 T
100
so 2 E c
so
I m .z I
40
z
20
0
FIG. 3. Spectra of the nucleosides (B) after they were oxidized, reduced, Materials and Methods.
from CN-Cbl and derivatized
(A) and D,-CN-Cbl as described under
662
SUNDIN
from the labeled nucleoside and represented 6.5% of the abundance of the 305 ion. This ion arises from a portion of the nucleoside that does not contain the benzene portion of the base rather than from contamination with unlabeled nucleoside because detectable (~1%) amounts of ions 245, 393,477, and 565 were not present in the spectrum of the labeled nucleoside. Figure 4 shows the proposed structure and fragmentation points of the two most abundant ions for the fully derivatized 5,6-dimethylbenzimidazole nucleoside of native Cbl. The larger 565 ion represents the fully derivatized, M-57 form of the nucleoside. This form represents a species in which a t-butyl fragment (atomic mass = 57) has been lost at any one of the positions x, y, or z. The smaller 303 ion represents a form of the nucleoside in which the ribose ring has been cleaved, with a fragment of the ribose ring remaining attached to the base portion of the nucleoside. It was this ribose fragment/base form (M-319) that was consistent and was monitored with the M-57 form. The M-57 and M-319 ions proved to be descriptive of each nucleoside, the only difference in atomic mass being attributed to variation in the base of the nucleoside. Figure 5 shows chromatograms of an experiment in which the nucleosides from CN-Cbl (top), [5(6)-OHBZAICN-Cba (middle), and the nucleoside fl-adenosine (bottom), the common form of adenosine, were subjected to hydrolysis conditions, oxidized with periodate, reduced with borohydride, derivatized, and analyzed by GC/MS. These three nucleosides represent all three ionic forms of nucleosides which result from Ce(OH)3 hydrolysis. As can be seen in Fig. 5, each nucleoside being analyzed eluted from the gas chromatography column at times that were clearly distinguishable from each other. We also applied our technique to a mixture of intact CN-Cbl and six naturally occurring bacterial Cbl analogues. Table I shows the predicted ions monitored and retention times for all seven nucleosides. It was clear from
(Y)
AND
ALLEN
1
6.0
7.0 Ion 303.20
60 Ion 667.40 G r E a
