Aerial oxidation of the glucocorticoid side-chain under pH control

Article abstract of DOI:10.1016/j.steroids.2005.08.003

The outcome of aerial oxidation of the glucocortico-steroid side-chain (as exemplified by dexamethasone) has been shown to be subject to strict pH control. At pH 7.4 the glyoxal is the only product; at pH values of 8 and 9.2 the etioacid is formed, and at

Full text of DOI:10.1016/j.steroids.2005.08.003

steroids 7 1 ( 2 0 0 6 ) 34–41
available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/steroids
Aerial oxidation of the glucocorticoid side-chain
under pH control
John S. Edmondsa,∗, Masatoshi Moritaa, Peter Turnerb, Brian W. Skeltonc,
Allan H. Whitec
a
Endocrine Disrupter Research Laboratory, National Institute for Environmental Studies, 16-2 Onogawa, Tsukuba,
Ibaraki 305-8506, Japan
b
School of Chemistry, University of Sydney, NSW 2006, Australia
Chemistry M313, University of Western Australia, Crawley, WA 6009, Australia
c
a r t i c l e i n f o
a b s t r a c t
Article history:
The outcome of aerial oxidation of the glucocortico-steroid side-chain (as exemplified by
dexamethasone) has been shown to be subject to strict pH control. At pH 7.4 the glyoxal is
the only product; at pH values of 8 and 9.2 the etioacid is formed, and at pH values of 13 or
above the epimeric glycolic acids are produced. The glycolic acid epimer that predominates
by a factor of 2 and is more stable has been shown by an X-ray crystal structural analysis
to be the 20R compound. The presence of arsenite changes the course of the reaction and
only the glycolic acids are yielded at pH values of 8 and above.
Received 21 June 2005
Received in revised form 27 July 2005
Accepted 2 August 2005
Available online 23 September 2005
Keywords:
© 2005 Elsevier Inc. All rights reserved.
Glucocorticoids
Dexamethasone
Aerial oxidation
NMR spectroscopy
Structure determination
1.
Introduction
Early workers established that oxidative treatment of 21-
hydroxy-20-ketosteroids yielded etioacids in the presence of
The transformation of the hydroxy-ketone sidechain of
glucocortico-steroid hormones and related steroids under
alkaline conditions is complex with products depending on
the presence or absence of oxygen, the presence or absence of
a 17-OH group and the nature of the base or solvent used [1–4].
One aspect of this complicated picture is that aerial oxidation,
or oxidation in the presence of oxygen, of the hydroxyketone
side-chain has been reported to produce both etioacids and
glycolic acids. Although the former were produced with dilute
alcoholic NaOH or KOH and the latter with aqueous base, it
was not clear to us what governed the reactions or why the
outcomes should be different.
air (or oxygen) and alkali [2,3]. The 21-acetates were stud-
ied and only enough alcoholic sodium or potassium hydrox-
ide to saponify the acetate and bring about the cleavage was
used (viz., two equivalents). There was no suggestion in their
reports of the production of glycolic acids. The reaction was
described as an oxidative cleavage and 1 mol of base and
1 mol of oxygen was required for the conversion of 1 mol of
21-hydroxy-20-ketosteroid to 1 mol of etioacid, with the con-
comitant generation of 1 mol of formate.
Lewbart and Mattox, on the other hand, reported that treat-
ment of the glyoxal 3␣-hydroxy-11,20-dioxo-5␤-pregnan-21-al
with aqueous sodium hydroxide yielded the epimeric pair of
∗
Corresponding author. Tel.: +81 29 850 2860; fax: +81 29 850 2870.
E-mail address: edmonds.john.s@nies.go.jp (J.S. Edmonds).
0039-128X/$ – see front matter © 2005 Elsevier Inc. All rights reserved.
doi:10.1016/j.steroids.2005.08.003

steroids 7 1 ( 2 0 0 6 ) 34–41
35
Scheme 1 – Transformations of dexamethasone (1) in aqueous buffers at elevated pH values and in the presence of arsenite.
The protons, methyl groups and side-chain carbons in 1 are numbered (see Table 1).
20-hydroxy acids [5]. The same compounds were also obtained
by cupric acetate oxidation of the glyoxal. Other papers have
reported on non-specific degradation of corticosteroids [6].
Here, we report the results of the oxidation of the hydroxy-
ketone sidechain of dexamethasone 1 (Scheme 1) in buffered
aqueous solutions at pH values above neutrality and show that
oxidation products are entirely dependent on pH. In addition
to this, the recent report [7] that arsenic, specifically arsenite
in very low doses, markedly affected glucocorticoid function
prompted us to determine any direct interactions between
arsenite and glucocorticoid hormones, as represented by 1.
Specifically we studied the effects of arsenite on the aerial oxi-
dation of the dexamethasone side-chain.
Meltings points were measured on an As-One ATM-01 melt-
ing point apparatus (As-One Co. Ltd., Osaka, Japan) and are
uncorrected.
2.2.
Buffers
Buffers (0.1 M) were made according to formulas listed in
the literature [8,9]. Buffers of pH values 7.4 and 8.0 were
made with phosphate, and those of 9.2 and 10.8 with carbon-
ate/bicarbonate. The same quantities of reagents (weights of
chemicals, volumes of solvent) were used for making buffers
in D2O as if they were being made in water. No consistent
differences were found between readings made for buffers in
D2O and those in water [Piccolo 2 HI 1290 m with amplified pH
electrode (Hanna Instruments, Portugal) and a ISFET KS723
solid state meter (Shindengen, Tokyo, Japan)] and the term pH
is used here for measurements made in both D2O and water
with no correction for isotope effects.
2.
Experimental
2.1.
General
1H NMR spectra were recorded in D2O on a JEOL ECA 800
spectrometer (JEOL, Tokyo, Japan) operating at 800 MHz. Shifts
are reported relative to external TMS. 1H assignments were
aided by 1H–1H COSY experiments. Fast atom bombardment
mass spectrometry (FAB MS) was carried out on a JEOL
JMS-700 GC/MS (JEOL, Tokyo, Japan). Benzoylformic acid, 2-
hydroxyacetophenone, and DCl and NaOD solutions were
obtained from the Aldrich Chemical Company, Tokyo, Japan.
Dexamethasone and all other chemicals and solvents were
obtained from Wako Pure Chemical Industries, Osaka, Japan.
2.3.
Structure determination
The available material was microcrystalline and, data being
insusceptible of access with a conventional charged-coupled
device area detector, recourse was made to synchrotron
instrumentation.
A full sphere of data was measured
(ꢀ = 0.550 A, 2ꢁmax = 42◦, T ca. 110 K) yielding 29,512 reflections,
these merging to 2637 unique after ‘empirical’/multiscan
‘absorption correction’ (ꢂ = 0.064 mm−1; ‘T’min/max = 0.90), 2353
being used in the full matrix least squares refinement,
˚

36
steroids 7 1 ( 2 0 0 6 ) 34–41
Fig. 1 – Molecular projection of dexamethasone glycolic acid 4.
anisotropic displacement parameters being refined for C, O,
F, (x, y, z, Uiso being constrained at estimates after location
tification of the products, further experiments were carried
out in NMR tubes and followed by 1H NMR spectroscopy to
refine information on the transformations observed. These
included transformations carried out in the presence of arsen-
ite. Sodium meta-arsenite solutions in D2O were prepared
immediately before use. Reactions to be followed by 1H NMR
spectroscopy were carried out at 25 ◦C in capped NMR tubes
(5 mm dia), no effort being made to exclude atmospheric oxy-
gen. Saturated solutions of 1 were made by ultrasonication of
an excess of the compound in the appropriate buffer in D2O
(2 ml) for 5 min. A portion (1 ml) of the solution was then fil-
tered into an NMR tube. This solution was assumed to be satu-
rated (i.e., 100 ␮g/ml, 0.25 mM [12]). Experiments in NMR tubes
were carried out with phosphate buffer (pH values of 7.4 and
8.0) and carbonate/bicarbonate buffer (pH values 9.2 and 10.8)
and in 0.05 M sodium deuteroxide in D2O. Titrations of chem-
ical shifts (ı values) against pH were carried out with acidic
compounds to provide additional information (Fig. 2a–c). pH
values of D2O solutions used in these titrations were adjusted
with NaOD and DCl solutions. Preliminary identification of
products in these NMR experiments were confirmed by the
following syntheses.
)
H
of the hydroxylic hydrogen atoms in difference maps. Conven-
tional residuals on |F| at convergence were R = 0.060, Rw = 0.080
(weights: (ꢃ2(F) + 0.0004 F2)−1). Neutral atom complex scatter-
ing factors were employed within the Xtal 3.7 program system
[10], chirality of the structure being assigned from the chem-
istry. A molecular projection is shown in Fig. 1 (50% probability
amplitude displacement ellipsoids for C, O, F, hydrogen atoms
˚
being shown with arbitrary radii of 0.1 A); a full .cif deposition
(excluding structure factor amplitudes) has been made with
the Cambridge Crystallographic Data Centre, #271377.
2.3.1. Crystal data
C
22H31FO7, M = 426.5. Orthorhombic, space group P212121
(D42, No. 19),
a = 6.022(5),
b = 12.731(5),
c = 25.946(5) A,
˚
V = 1989 A . Dc (Z = 4) = 1.424 g cm−3
.
3
˚
2.3.2. Comment
The conformation and geometry of the polycyclic skeleton are
essentially similar to those of dexamethasone [11].
The compound crystallizes as a monohydrate, in a struc-
ture which displays widespread intra- and inter-molecular
hydrogen-bonding from/between the hydroxylic and car-
bonyl entities (but not the fluorine): HO(11). . .O(3)(2 − x, ½+ y,
½− z) 2.0; HO(17). . .O(20) 2.2; HO(20). . .O(22)(x − ½, 1½− y,
1 − z) 2.1; HO(22). . .O(w)(1 + x, y, z) 1.7; H(wa). . .O(21) 1.9;
2.4.1. Dexamethasone glyoxal (2)
A solution of 1 (50 mg, 0.13 mmol) in phosphate buffer (500 ml,
0.1 M, pH 7.4) was left at 25 ◦C for 28 days. After this time the
1H NMR spectrum showed that 1 had been completely con-
verted to a single compound. The solution was evaporated to
dryness (<40 ◦C) and the residue extracted with methanol. The
methanol extract was then subjected to size-exclusion chro-
matography (SEC) on Sephadex LH-20 (elution with methanol)
and dexamethasone glyoxal (2, 33 mg, 66%) obtained as a
white crystalline powder (Mp 126 ◦C, lit. 127–130 [13]). 1H NMR
(Table 1); FAB + MS (methanol solution) m/z 423 (hemiacetal,
M + 1).
˚
H(wb). . .O(3)(x − 1, y + 1, z) 1.9 A (all est.).
2.4.
Reactions in NMR tubes
Reactions were initially carried out in NMR tubes and were
mainly followed by observation of the 16-methyl signals. Fol-
lowing this, reactions were carried on a sufficient scale to
provide product compounds for characterization. After iden-

steroids 7 1 ( 2 0 0 6 ) 34–41
37
Fig. 2 – Plots of chemical shift (ı) against pH for (a) 16-methyl group protons, (b) 18-methyl group protons, and (c) protons on
C20. Squares, etio acid, compound 3; triangles, glycolic acid, compound 4; circles, glycolic acid, compound 5.
2.4.2. Dexamethasone etioacid (3)
2.4.3. Dexamethasone epimeric glycolic acids (4 and 5)
A solution of 1 (50 mg, 0.13 mmol) in carbonate/bicarbonate
buffer (500 ml, 0.1 M, pH 9.2) was left to stand at room tem-
perature overnight. The solution was then evaporated to dry-
ness (<40 ◦C), the residue extracted with methanol and sub-
jected to SEC as for glyoxal 2. Dexamethasone etioacid 3 was
obtained as a white powder (34 mg, 71%; recrystallized from
acetone/light petroleum to give colorless needles, Mp >260 ◦C
decomp., lit. >258 ◦C decomp. [14]). 1H NMR (Table 1); FAB-MS
m/z (M − 1) 377.
A solution of 1 (100 mg, 0.25 mol) in 0.05 M aqueous sodium
hydroxide (1 l) was left at room temperature overnight. The
pH of the solution was reduced to 7 (HCl) followed by evap-
oration to dryness (<40 ◦C). The residue was subjected to
fractional crystallization (aqueous methanol) and the least
soluble epimer 4 obtained as colorless needles (37 mg, Mp
254–256 ◦C); the more soluble epimer 5 obtained as small color-
less prisms (7 mg, Mp >300 ◦C). NMR data for both compounds
are given in Table 1. High resolution FAB + MS, compound 4,

Table 1 – 1H NMR data for dexamethasone (1) and its derivatives
Proton
Compound
Dexamethasone
Dexamethasone, 1,
Dexamethasone
glyoxal, 2, D2O
Dexamethasone
etioacid, 3, D2O
unbuffered
Dexamethasone
glycolic acid, 5,
methanol-d4
Dexamethasone 17-ketones
D2O
glycolic acid, 4,
methanol-d4
1
2
7.543, d, J1,2 = 10.1 Hz
6.437, dd, J1,2 = 10.1 Hz,
J2,4 = 1.8 Hz
7.546, d, J1,2 = 11.0 Hz
7.568, d, J1,2 = 10.1 Hz
6.432, dd, J1,2 = 10.1 Hz,
J2,4 = 1.8 Hz
7.419, d, J1,2 = 10.1 Hz
6.265, dd, J1,2 = 10.1 Hz,
J2,4 = 1.8 Hz
7.419, d, J1,2 = 10.1 Hz
6.442,
J1,2 = 11.0 Hz,
6.267,
dd,
J2,4 = 1.8 Hz
J1,2 = 10.1 Hz,
J2,4 = 1.8 Hz
4
6A
6.244, d, J2,4 = 1.8 Hz
2.487
6.247, d, J2,4 = 1.8 Hz
2.478
6.239, d, J2,4 = 1.8 Hz
2.486
6.065, d, J2,4 = 1.8 Hz
2.385
6.059, d, J2,4 = 1.8 Hz
2.381
6B
2.735
2.738
2.742
2.706
2.699
2.765
2.765
7A
1.490
1.490
1.489
1.498
1.512
7B
1.985
1.989
1.990
1.880
1.868
8
2.564
2.578
2.513
2.413
2.411
11
4.410
4.396
4.365
4.168
4.170
12A
12B
14
1.564
2.188
2.115
1.615
2.199
2.174
1.511
2.059
1.956
1.639
2.015
2.061
1.747
2.073
2.008
15A
15B
16
1.289
1.843
3.049
1.302
1.858
3.068
1.232
1.808
2.975
1.127
1.705
2.483
1.156
1.668
2.558
2.611
2.152
16Me
18Me
19Me
20
0.889, d, 7.3 Hz
0.989
1.567
0.894, d, 7.3 Hz
1.024
1.566
0.885, d, 7.3 Hz
1.093
1.576
1.074, d, 7.3 Hz
1.176
1.578
0.918, d, 7.3 Hz
1.176
1.584
1.059, d, 7.4 Hz
1.139
1.613
1.209, d, 7.4 Hz
1.233
1.613
4.180
3.925
21A
21B
4.687, d, JA,B = 18.7 Hz
4.416, d, JA,B = 18.7 Hz

steroids 7 1 ( 2 0 0 6 ) 34–41
39
m/z (M + 1) 409.2013, required for C22H30FO6 409.2027; com-
pound 5, m/z (M + 1) 409.2015, required for C22 30FO6 409.2027.
3.
Results and discussion
H
Also isolated by SEC from the liquors remaining after crys-
tallization were 9-fluoro-11␤-16␣-methylandrosta-1,4-diene-
3,17-dione and 9-fluoro-11␤-16␤-methylandrosta-1,4-diene-
3,17-dione; 1H NMR (Table 1). FAB-MS m/z (M − 1) 331; FAB + MS
333.
Aerial oxidation of 1 in aqueous solution at pH 7.4 was slow
(weeks) and yielded the glyoxal (Scheme 1, 2); the reaction did
not proceed further. At pH values of 8 and 9.2 the etioacid
3 was the only product of oxidation of both 1 and 2. At pH
values in excess of 12 the glycolic acids (4) and (5) were the
only products of the oxidation of 1. An NaOH concentration of
0.05 M was sufficient for the reaction to be completed in 0.5 h.
As these reactions were followed by monitoring the 1H NMR
spectra, in particular of the 16- and 18-Me groups of 1 and
its transformation products, we carried out a full examination
of the dependence of the chemical shifts of the resonances
attributed to these groups on pH. The results are plotted in
Fig. 2a for the 16-methyl groups on the etio acid 3 and the gly-
colic acids 4 and 5, and in 2b for the 18-methyl groups on the
three compounds. In addition, the chemical shifts of the pro-
tons at C20 were diagnostic in the identification of compounds
4 and 5, and Fig. 2c shows the dependence of the chemical shift
of these protons on pH for the two compounds. The approxi-
mate pK values (typical of carboxylic acids at around 5) can be
read from the plots.
2.4.4. Dexamethasone (1) to epimeric glycolic acids (4,5)
in the presence of arsenite
Sodium meta-arsenite (5 ␮l of 0.5 M solution in D2O) was added
to a saturated solution of 1 in carbonate/bicarbonate buffer
in D2O (1 ml, 0.1 M, pH 9.2) and the sample transferred to
an NMR tube. The arsenite concentration was 2.5 mM, 10
times that of the steroid. 1H NMR spectra were recorded
at intervals appropriate to the rate of transformation of
1. The pH value of 9.2 represents a typical reaction but
essentially similar results were obtained in the pH range of
8–10.8.
2.4.5. Dexamethasone glyoxal (2) to etioacid (3)
A saturated solution of 2 in carbonate/bicarbonate buffer
in D2O (1 ml, pH 9.2) was placed in an NMR tube and the
conversion of the glyoxal to the etioacid monitored by 1H
NMR spectrsocopy. Dissolution with a pH of 8 gave a similar
result.
The two glycolic acid epimers (4) and (5) were not produced
in equal quantity; the major product contributed about 65%
of the total (see below). Exposure of an aqueous solution of
1 to a pH of 11 yielded a mixture of both the etioacid 3 and
the glycolic acids 4 and 5 with the latter compounds again
appearing in a 2:1 ratio.
2.4.6. Dexamethasone glyoxal (2) to glycolic acids (4,5)
Compound 2 (about 100 ␮g) was dissolved in 0.05 M NaOD in
D2O and the conversion of the glyoxal to the epimeric glycolic
acids monitored by 1H NMR spectroscopy.
It was proposed [5] that the conversion of hydroxyketone
(via the glyoxal) to the epimeric pair of glycolic acids involved
an intramolecular Cannizzaro reaction. Support for this pro-
posal came from the observation that 17-OH steroids, in which
17(20) enolization is not possible, underwent the reaction at a
greater rate and under milder conditions (1.25 M alkali, 0.5 h)
than 17-deoxy compounds (4 M alkali, 8 h). We have found
that 1, in saturated solution in 0.05 M aqueous NaOH, is fully
converted to the epimeric glycolic acids 4 and 5 in 0.5 h.
Although these are mild conditions for a Cannizzaro reac-
tion, evidence for the essential hydride shift was provided
by carrying out the reaction in D2O/NaOD. That the C20s in
the new compounds bore hydrogens rather than deuteriums
was evident from the 1H NMR spectra. The involvement of
a Cannizzaro-type reaction would require that the first stage
to be the conversion of the hydroxyketone steroid to its gly-
oxal and, indeed, 2 (made by aerial oxidation of 1 at pH 7.4)
yielded 4 and 5 under base treatment at the same rate as did 1
itself.
The epimers 4 and 5, synthesized on a preparative scale in
0.05 M NaOH, were separated by fractional crystallisation and
an X-ray structure analysis showed that the major compound
4 had the stereochemistry shown in Scheme 1 and Fig. 1 (20R).
Simple modelling indicated that formation of the 20R com-
pound was likely to be favoured over the 20S epimer because
of steric factors. Isolation and crystallization of the minor gly-
colic acid 5 was problematical; the compound was apparently
labile—much more so than its epimer 4—with degradation by
loss of the side-chain to yield further compounds, provision-
ally identified as 9-fluoro-11␤-16␣-methylandrosta-1,4-diene-
3,17-dione and 9-fluoro-11␤-16␤-methylandrosta-1,4-diene-
2.4.7. Dexamethasone glyoxal (2) to glycolic acids (4,5) in
the presence of arsenite
Sodium meta-arsenite (5 ␮l of 0.5 M solution in D2O) was added
to a saturated solution of 2 in carbonate/bicarbonate buffer
in D2O (1 ml, pH 9.2). the arsenite concentration was 2.5 mM,
approximately 10 times that of the steroid. The solution was
tranferred to an NMR tube and the conversion of the glyoxal
to the glycolic acids monitored by 1H NMR spectroscopy. Solu-
tions with pH values of 8 and 10.8 gave essentially the same
results.
2.4.8. 2-Hydroxyacetophenone with sodium deuteroxide
The reactions of a solution of 2-hydroxyacetophenone (0.5 mg)
in 0.05 M NaOD in D2O (1 ml) were monitored by 1H NMR
spectroscopy. Signals for 2-hydroxyacetophenone were dimin-
ished with the concomitant appearance of peaks read-
ily attributable to benzoic acid and mandelic acid. Fifty
percent of the 2-hydroxyacetophenone had been trans-
formed after 10 h at 25 ◦C. About 70% of the trans-
formed 2-hydroxyacetophenone was converted to benzoic
acid and 30% to mandelic acid. Reactions carried out in
the dark and in normal laboratory lighting gave the same
results.
2.4.9. Benzoylformic acid with sodium deuteroxide
A solution of benzoylformic acid (0.5 mg) in 0.05 M NaOD in
D2O (1 ml) was monitored by 1H NMR spectroscopy. No reac-
tion was observed.

40
steroids 7 1 ( 2 0 0 6 ) 34–41
3,17-dione (by 1H NMR and FAB MS of an unseparated mixture
and by analogy with the 3(␣),17(␣),21-trihydroxypregnane-
11,20-dione series [4]). These 17-ketones have previously been
identified as impurities in commercial samples of dexam-
ethasone sodium phosphate [15]. It was not clear to us why
this should occur under the reaction conditions employed,
although it has been reported that the hydroxyketone side-
chain of glucocorticoid steroids undergo retrograde aldol reac-
tions to yield such ketones when treated with alkali in the
absence of air. Nor was it clear why the 20S epimer should
be so much less stable than its 20R epimer. It was previously
noted that the formation of the methyl esters of the anal-
ogous glycolic acids (by cupric acetate in methanol) in the
prednisolone series favours the formation of one epimer over
the other [16]. However, those authors reported, with no obvi-
ous basis, that the 20S compound predominated [16]. A later
publication from the same group described the conversion of
1, as well as prednisolone, to methyl esters of their glycolic
acids [17]. Both epimers were synthesized but only the 20S
compound was further reacted; it was hydrolyzed to the free
acid with methanolic KOH. Possibly the epimer considered to
be the 20R compound was omitted from this step because of its
lability. Again no reason was given for the assignment of the
stereochemistry of the 20 carbon. We believe, on the basis of
the X-ray crystal structural determination of 4 that the stere-
ochemical assignments [16,17] in both the prednisolone and
dexamethasone series need revision; the predominant and
stable epimer in the dexamethasone series is the 20R com-
pound 4 and it likely to be so in the prednisolone series also.
With regard to the stability of steroidal glycolic acids of this
type and their methyl esters, it is also noteworthy that Lewbart
and Mattox [5] made no mention of any instability of either
of the epimeric 20-hydroxyacids derived from 3␣-hydroxy-
11,20-dioxo-5␤-pregnan-21-al (i.e., an analogous compound
but lacking a 17-hydroxyl group) by treatment with aqueous
NaOH.
values of 8–11 caused the course of the oxidation to change
completely, the glycolic acids 4 and 5 being the only products;
the etioacid 3 was not produced at all.
In order to explain the manner in which arsenite subverts
the course of the oxidative cleavage reaction that yielded 3,
we sought a more detailed explanation of its mechanism.
Several possibilities were considered including the decarboxy-
lation or decarbonylation of an intermediate ketoacid and a
photochemical cleavage. 2-Hydroxyacetophenone served as
a model compound for this conversion but was rather less
reactive than the steroid; at pH values of 12 or less reac-
tion was very slow and at pH 13 or higher it yielded about
65% benzoic acid and 35% mandelic acid with reaction com-
plete in a few hours at room temperature (Scheme 2), offer-
ing parallels, then, with the steroid conversion to both the
etioacid and the glycolic acids. However, benzoylformic acid
proved to be stable in aqueous solution at pH values in
excess of 13 (Scheme 2) and therefore could not be an inter-
mediate in the transformation of 2-hydroxyacetophenone to
benzoic acid. Thus the possibility of a ketoacid intermedi-
ate in the generation of the etioacid was eliminated. We
then investigated the possible reaction of singlet oxygen with
the double bond of the hydroxy-keto side-chain in enolate
form, and considered whether the enolate might function
as a photosensitizer and possibly provide an adequate chro-
mophore for laboratory lighting. However, the conversion of
2-hydroxyacetophenone to benzoic acid/mandelic acid at pH
13+ occurred at the same rate in the dark as in normal
laboratory light, so the involvement of singlet oxygen was
also shown to be unlikely. Details of the mechanism remain
obscure.
How then does arsenite cause the production of glycolic
acids rather than the etioacid at pH values of 8 and 9.2? We
considered the possibility that the oxidation of arsenite to
arsenate exhausts the available oxygen in the capped NMR
tubes in which the reactions were carried out, and thereby
renders the production of the etioacid (which requires oxy-
gen) impossible, and opens the pathway to the glycolic acids.
However the reported [18] rate of oxidation of of arsenite to
We next turned our attention to possible effects of arsenite
on these transformations and found that the addition of a 10
times molar excess of arsenite to aqueous solutions of 1 at pH
Scheme 2 – The reactions of benzoylformic acid and 2-hydroxyacetophenone with aqueous sodium deuteroxide.

steroids 7 1 ( 2 0 0 6 ) 34–41
41
[4] Wendler NL, Graber RP. Alkaline degradation of the
arsenate at the relevant pH values would be much too slow
to support such an explanation. Possibly arsenite complexes
in some way with the oxygen functions of the glucocorti-
coid sidechain and promotes the reaction to the glycolic acids
and/or hinders the production of the etioacid. There are two
observations that may be relevant. First, at pHs 8 and 9.2 in
the presence of arsenite the glycolic acids 4 and 5 are pro-
duced in exactly equal amounts (as opposed to the 2 to 1 ratio
observed when 1 was subjected to pH 13+). And second, sub-
jection of the glyoxal 2 to arsenite at pH 9.2 produced 4 and 5
at a slower rate than did 1 itself when subjected to the same
conditions. This implies that, in the presence of arsenite at
pH values of 8 and 9.2, the route to the glycolic acids is not
through its glyoxal and raises the unexpected possibility that
the mechanism for the production of 4 and 5 from 1 at pH 13+,
and at pH values 8 and 9.2 in the presence of arsenite, might
be different – through its glyoxal in the former case but not
in the latter. However, when the reaction in the presence of
arsenite is carried out in D2O a proton rather than a deuteron
is still gained at C20, so that a hydride shift is involved in the
transformation.
cortical side chain. Chem Ind 1956:549.
[5] Lewbart ML, Mattox VR. Conversion of steroid-17-yl
glyoxals to epimeric glycolic esters. J Org Chem
1963;28:1779–86.
[6] Dekker D, Beijnen JH. The C-16 methyl group orientation
influences alkaline anaerobic decomposition of the
dihydroxyacetone moiety of corticosteroids. Acta Pharm
Suec 1981;18:185–92, and references cited therein.
[7] Kaltreider RC, Davis AM, Lariviere JP, Hamilton JW. Arsenic
alters the function of the glucocorticoid receptor as a
transcription factor. Environ Health Perspect
2000;109:245–51.
[8] Gomori G. Preparation of buffers for use in enzyme
studies. In: Colowick SP, Kaplan NO, editors. Methods in
Enzymology. 1955;1:138–46.
[9] Delory GE, King EJ. A sodium carbonate-bicarbonate buffer
for alkaline phosphatases. Biochemistry 1945;39:245.
[10] Hall SR, du Boulay DJ, Olthof-Hazecamp R, editors. The
Xtal 3.7 System. University of Western Australia; 2001.
[11] (a) van den Bossche G. Crystallographic study of some
steroids. I. 9␣-Fluorocortisol dexamethasone, and
6␣-methylprednisolone. II. Estrone and estradiol. Bull Soc
R Sci Liege 1971;40:614–27 (CCDC DEXMETOI);
(b) Rohrer DG, Duax WD. 9␣-Fluoro-16␣-methyl-11␤,17␣,21-
trihydroxy-1,4-pregnadiene-3,20-dione. Cryst Struct
Commun 1977;6:123–6.
[12] The Merck Index, 11th ed., Merck and Co., Inc., Rahway,
N.J. USA, 1989, p. 464.
[13] Marx M, Leonia NJ, Smith LL, inventors; American
Cyanamid Co., assignee. 21-Nitrogen substituted steroids
of the pregnane series and methods of preparing the
same. US Patent 3 020 275. 1962, Feb 6.
Acknowledgements
We thank Ms. M. Katsu for assistance. Use of the Chem-
MatCARS Sector 15 at the Advanced Photon Source, was
supported by the Australian Synchrotron Research Program,
which is funded by the Commonwealth of Australia under the
Major National Research Facilities Program. ChemMatCARS
Sector 15 is also supported by the National Science Founda-
tion/Department of Energy under grant numbers CHE9522232
and CHE0087817 and by the Illinois Board of Higher Education.
The Advanced Photon Source is supported by the U.S. Depart-
ment of Energy, Basic Energy Sciences, Office of Science, under
Contract No. W-31-109-Eng-38.
[14] Phillipps GH, May PJ, inventors; Glaxo Laboratories Ltd.,
UK, assignee. Derivatives of
17␣-hydroxy-androst-4-17␤-carboxylic acids. UK Patent 1
384 372. 1975, Feb 19.
[15] Juenge EC, Brower JF. High-performance liquid
chromatographic separation and identification of epimeric
17-ketone impurities in commercial sample of
dexamethasone sodium phosphate. J Pharm Sci
1979;68:551–4.
[16] Kimura T, Yamaguchi T, Usuki K, Kurosaki Y, Nakayama T,
Fujiwara Y, et al. Colonic mucosa-specific “pro-antedrugs”
for oral treatment of ulcerative colitis: design, synthesis
and fate of methyl 20-glucopyranosyloxyprednisolonates. J
Control Release 1994;30:125–35.
r e f e r e n c e s
[1] Mason HL. Chemical studies of the suprarenal cortex. J
Biol Chem 1938;124:475–9.
[17] Suzuki T, Sato E, Tada H, Tojima Y. Examination of local
anti-inflammatory activities of new steroids, hemisuccinyl
methyl glycolates. Biol Pharm Bull 1999;22:816–21.
[18] Kuenhelt D, Goessler W, Irgolic KJ. The oxidation of
arsenite in aqueous solutions. In: Abernathy CO, Calderon
RL, Chappell WR, editors. Arsenic exposure and health
effects. London: Chapman and Hall; 1997. p. 45–54.
[2] Velluz L, Petit A, Pesez M, Beret R. Saponification de
l’ace´tate de de´soxycorticoste´rone. Bull Soc Chim France
1947:123–6.
[3] Herzig PT, Ehrenstein M. Investigations on steroids. XIV.
Studies in the series of the adrenal cortical hormones:
dehydration of 6␤,21-diacetoxyallopregnane-5-ol-3,20-
dione. J Org Chem 1951;16:1050–63.