One-Step Chemo-, Regio- and Stereoselective Reduction of Ketosteroids to Hydroxysteroids over Zr-Containing MOF-808 Metal-Organic Frameworks

Article abstract of DOI:10.1002/chem.202100967

Zr-containing MOF-808 is a very promising heterogeneous catalyst for the selective reduction of ketosteroids to the corresponding hydroxysteroids through a Meerwein-Ponndorf-Verley (MPV) reaction. Interestingly, the process leads to the diastereoselective synthesis of elusive 17α-hydroxy derivatives in one step, whereas most chemical and biological transformations produce the 17β-OH compounds, or they require several additional steps to convert 17β-OH into 17α-OH by inverting the configuration of the 17 center. Moreover, MOF-808 is found to be stable and reusable; it is also chemoselective (only keto groups are reduced, even in the presence of other reducible groups such as C=C bonds) and regioselective (in 3,17-diketosteroids only the keto group in position 17 is reduced, while the 3-keto group remains almost intact). The kinetic rate constant and thermodynamic parameters of estrone reduction to estradiol have been obtained by a detailed temperature-dependent kinetic analysis. The results evidence a major contribution of the entropic term, thus suggesting that the diastereoselectivity of the process is controlled by the confinement of the reaction inside the MOF cavities, where the Zr⁴⁺ active sites are located.

Full text of DOI:10.1002/chem.202100967

Accepted Article
Accepted Article
Title: One-step Chemo-, Regio- and Stereoselective Reduction of
Ketosteroids to Hydroxysteroids over Zr-containing MOF-808
Metal-organic Frameworks
Authors: Hans-Hilmar Mautschke and Francesc X. Llabrés i Xamena
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: Chem. Eur. J. 10.1002/chem.202100967
Link to VoR: https://doi.org/10.1002/chem.202100967
01/2020

10.1002/chem.202100967
Chemistry - A European Journal
FULL PAPER
One-step Chemo-, Regio- and Stereoselective Reduction of
Ketosteroids to Hydroxysteroids over Zr-containing MOF-808
Metal-organic Frameworks
H. –H. Mautschke,[a] and F. X. Llabrés i Xamena*[a]
[a] Dr. H. –H. Mautschke and Dr. F. X. Llabrés i Xamena
Instituto de Tecnología Química, Universitat Politècnica de València, Consejo Superior de Investigaciones Científicas,
Avda. de los Naranjos s/n, 46022 Valencia, Spain.
E-mail: fllabres@itq.upv.es
Twitter handle: @ITQ_UPVCSIC
Supporting information for this article is given via a link at the end of the document.
Abstract: Zr-containing MOF-808 is a very promising heterogeneous
catalyst for the selective reduction of ketosteroids to the
corresponding hydroxysteroids through a Meerwein-Ponndorf-Verley
(MPV) reaction. Interestingly, the process leads to the
diasteroselective synthesis of elusive 17α-hydroxy derivatives in one
step, while most chemical and biological transformations produce the
17β-OH compounds, or they require several additional steps to
convert 17β-OH into 17α-OH by inverting the configuration of the 17
center. Moreover, MOF-808 is found to be stable and reusable, and it
is also chemoselective (only keto groups are reduced, even in the
presence of other reducible groups such as C=C bonds) and
regioselective (in 3,17-diketosteroids only the keto group in position
17 is reduced, while the 3-keto group remains almost intact). Kinetic
rate constant and thermodynamic parameters of estrone reduction to
estradiol have been obtained by a detailed temperature-dependent
kinetic analysis. The results evidence a major contribution of the
entropic term, thus suggesting that the diastereoselectivity of the
process is controlled by the confinement of the reaction inside the
MOF cavities, where the Zr4+ active sites are located.
synthetic analogs used in medicine, such as glucocorticoids and
corticoids in general.[2] Meanwhile, hydroxysteroids are also
relevant compounds in the pharmaceutical industry, and in
particular those containing a hydroxyl group attached to C17,
such as estradiol, androstanediol or testosterone/epitestosterone.
One possible route to prepare these hydroxysteroids is by a direct
stereoselective reduction of the corresponding ketosteroid, which
can be carried out by using either a variety of microorganisms[3]
or with chemical reducing agents.
Introduction
Scheme 1. Steroids skeleton and conventional atom and ring
labeling.
Steroids are a large group of chemical substances sharing a 17-
carbon atom skeleton composed of four fused rings (three six-
membered rings and one five-membered ring), conventionally
denoted A-D, as shown in Scheme 1.[1] Steroid compounds vary
from each other in the type of groups attached to this skeleton,
their position, and configuration. Small modifications in the
structure of these steroids produce significant differences in their
biological activity. Therefore, the development of novel routes to
modify existing steroids in a controlled way to improve their
activity or to prepare new steroid derivatives with potential
pharmacologic activity is an area of active research. In this sense,
chemo-, regio- and stereoselective (bio)transformation of steroids
using highly selective catalysts or microorganisms is crucial.
Within the large family of steroids, ketosteroids (also denoted
oxosteroids) are steroids in which at least one hydrogen atom of
the steroid nucleus has been replaced by a keto group (C=O).
Many endogenous hormones are ketosteroids, as well as their
In general, biotransformation has several advantages over
chemical routes, including regio- stereo- as well as
enantioselectivity, thus allowing the preparation of chiral products.
From an economic point of view, biotransformations can also be
cheaper than chemical methods. However, in many cases, the
reduction of ketosteroids to hydroxysteroids by microorganisms is
accompanied by unwanted side reactions, such as C=C bond
reduction or hydroxylation at other positions of the steroid
skeleton,[4–6] that decrease the final yield of the target
hydroxysteroids and complicate their purification and isolation.
Additional complications may arise in biocatalytic processes due
to the low solubility in aqueous medium of steroids in general, or
related to the regeneration of the cofactors (NADH or NADPH)
used as the source of hydride ions in the process.
1
This article is protected by copyright. All rights reserved.

10.1002/chem.202100967
Chemistry - A European Journal
FULL PAPER
Besides biocatalytic methods, reduction of ketosteroids can
be also achieved by using chemical reducing agents, such as
NaBH4, Zn(BH4)2 or LiAlH4, thought these methods present their
own limitations as well. For instance, reduction of 17-ketosteroids
with NaBH4 produces exclusively 17β-OH isomers due to the
steric hindrance of the 18-methyl group that blocks hydride attack
from the upper face (see Scheme 2). Therefore, when the product
of interest is the 17α-OH isomer, most efforts have been focused
on inverting the configuration of 17β-OH through Mitsunobu
reactions,[7,8] or by the formation and displacement of sulfonyl
ester derivatives.[9–11] However, both methods introduce further
synthetic steps, including protection/deprotection reactions using
expensive or non-commercial reagents, and complicating product
isolation and purification. As a result, the final yield obtained of
the target 17α-OH is usually very low.
mentioning that 17α-hydroxy-estradiol was obtained directly from
estrone in a single reaction step, using only isopropanol as the
sole reagent and avoiding any additional protection/deprotection
steps (as required, for instance, in Mitsunobu type processes[7,8]).
Encouraged by these promising results, we wanted to explore
in more detail the diastereoselective reduction of estrone over
MOF-808 materials. We have extended our studies to the
synthesis of other challenging 17α-hydroxysteroids of interest
from the corresponding 17-ketosteroids. Selection of these
compounds allowed us to further address other relevant aspects
of the catalytic process, such as regio- and chemoselectivity of
the MPV reaction as well as catalyst stability and reusability.
Results and Discussion
Diastereoselective reduction of estrone
The reduction of estrone (hereafter E1) at the 17 position can
yield a mixture of two estradiols, as shown in Scheme 3. These
two alcohols are diastereoisomers and are termed 17α- and 17β-
estradiol (hereafter α-E2 and β-E2 for short), depending on the
configuration of the 17-OH group relative to the 18-methyl group.
Scheme 2. Reduction of 17-ketosteroids by most chemical
reducing compounds yields 17β-OH compounds due to the steric
hindrance of the 18-methyl group.
Besides stereoselectivity, regioselectivity can also be an issue
when chemical reducing agents are used for the reduction of
steroids bearing more than one keto group. For instance, 3-
ketosteroids are in general more reactive than 17-or 20-
ketosteroids.[12] Therefore, the preparation of 17-OH products can
be challenging when more than one keto groups are present in
the starting ketosteroid, so that mixtures of various mono- and
polyhydroxylated products can be obtained. Finally, also
chemoselectivity of the reaction must be taken into account when
working with ketosteroids containing other reducible groups (such
as C=C bonds) besides the target carbonyl group.
Scheme 3. Reduction of E1 yields a mixture of α-E2 and β-E2
diastereoisomers.
As shown in Table 1 (entry 1), when reduction of E1 was
carried out over MOF-808 in isopropanol (iPrOH) at 393 K, the
reaction was almost complete after 8 h (97% conversion). The
corresponding turnover frequency (TOF), calculated at short
reaction time, was 2.44 h-1. As compared to iPrOH, the reaction
was slightly slower when 2-butanol (2-BuOH) was used (TOF =
1.33 h-1), but it eventually reached almost full conversion (91%)
after 8 h, and complete conversion after 24 h of reaction (entry 2).
The reaction also proceeded smoothly with 2-pentanol (2-
PentOH) and 1-phenyl ethanol (entries 3 and 4). In all cases, the
reduction of estrone produced only a mixture of α-E2 and β-E2,
and no other byproducts were detected (100% selectivity).
The results obtained with MOF-808 are in sharp contrast with
the little or no catalytic activity observed under the same reaction
conditions with other Zr-containing compounds (entries 5-8), or
using well-known MPV catalysts, such as Zr- and Al-
isopropoxides (entries 9-10).
Zirconium trimesate MOF-808 is a metal-organic framework
(see structure in Figure S1) that was first described in 2014 by
Furukawa et al.[13] This compound has recognized catalytic
activity for the Meerwein-Ponndorf-Verley (MPV) reduction of
various carbonyl compounds.[14–16] In this sense, we have
shown[16] that pristine and defect-engineered MOF-808 are more
active for this reaction than the archetypal zirconium terephthalate,
UiO-66.[17] This was attributed to
a high availability of
coordinatively unsaturated Zr4+ sites (cus) in MOF-808 upon
removal of formate ions used during the synthesis. Meanwhile,
cus in UiO-66 are only located at missing-linker defect sites and
are thus much less abundant.[18] A further advantage of MOF-808
compared to UiO-66 is its wider pore structure (pore apertures are
14 Å in MOF-808 and ~ 6 Å in UiO-66), which allows converting
bulkier substrates. We showed this by comparing the MPV
reduction of estrone, with approximate dimensions: 11.2 Å × 6.2
Å × 4.2 Å, using either MOF-808 or UiO-66 as catalyst. Our results
showed that UiO-66 was basically inactive for this reaction, while
MOF-808 produced the expected hydroxylated derivative
quantitatively. Interestingly, MPV reduction of estrone with
isopropanol over MOF-808 yielded the elusive 17α-estradiol with
moderate diastereoselectivity. Although the diastereoselectivity
attained in that work was still far from optimum, it is worth
The diastereomeric ratio of the two alcohols formed, α-E2:β-
E2, was determined at the end of the reaction from the
corresponding 1H NMR spectra of the reaction filtrate. The
procedure used is detailed in the Supporting Information. The
results obtained are also included in the last column in Table 1.
2
This article is protected by copyright. All rights reserved.

10.1002/chem.202100967
Chemistry - A European Journal
FULL PAPER
X receptor and the activation of MAPK/ERK.[19] α-E2 is of
pharmaceutical interest for treating degenerative diseases, such
as Alzheimer, Parkinson, Friedreich’s ataxia or stroke, with much
less secondary effects than the β-E2 isomer.[20–23] Therefore, the
development of new synthetic (catalytic) routes for the
preparation of this compound from cheap starting compounds,
such as estrone, has an obvious interest.
In our previous work,[16] a ca. 60:40 d.r. was achieved when
iPrOH was used at 353 K. Thus, although α-E2 was still the minor
product of the reaction, the amount obtained was still considerably
higher than in previous reports on the synthesis of similar 17α-OH
compounds.[12,24,25] In the present case, the results obtained with
iPrOH at 393 K were significantly better: 60% selectivity to α-E2
(entry 1), thus becoming the major product of the reaction. Since
the diastereoselectivity attained in the MPV reduction of estrone
does not seem to depend on the reaction temperature (see below,
Figure 2), we attribute the higher selectivity to α-E2 attained in the
present work to improvements in the MOF-808 preparation and
activation procedure (viz. formate removal) compared to our
previous study.
Still better results were obtained when 2-BuOH was used as
a solvent and reducing alcohol, attaining an impressive 87:13 d.r.
at 91% conversion after 8 h of reaction (entry 2). When the size
of the secondary alcohol used was further increased (viz., 2-
PentOH and 1-phenyl ethanol), the d.r. decreased to 73:28 and
60:40, respectively, thus evidencing that 2-BuOH provided the
optimum diastereoselectivity among the alcohols tested.
We thus considered the reaction with 2-BuOH at 393 K to
evaluate the stability and reusability of MOF-808. According to the
XRD characterization of the catalyst recovered after the MPV
reaction, the solid was found to be stable under the reaction
conditions used, and both the catalytic activity and
diastereoselectivity to α-E2 was maintained for at least 6 catalytic
cycles, as shown in Figure 1. Meanwhile, ICP analysis of the
filtrate after the reaction revealed a minimal Zr4+ leaching (less
than 1% of the total zirconium used) from the catalyst to the
reaction medium.
Table 1. Summary of the catalytic results for the MPV reduction of estrone over
various catalysts. [a]
Conv.
TOF[c]
(h-1)
2.44
Entry
1
Catalyst
Alcohol
iPrOH
d.r.[d]
(time)[b]
MOF-808
97% (8
h)
60:40
> 99%
(24 h)
91%
(8h)
2
MOF-808
2-BuOH
1.33
87:13
> 99%
(24 h)
92% (8
h)
97% (8
h)
2% (24
h)
13%
(24 h)
2% (24
h)
2% (24
h)
2% (24
h)
11%
(24 h)
99%
3
4
MOF-808
MOF-808
ZrO2
2-PentOH
1-phenylethanol
2-BuOH
2.11
73:28
60:40
44:56
50:50
65:35
52:48
50:50
34:66
3:97
2.77
5
-
-
-
-
-
-
-
6
ZrCl4
2-BuOH
7
Zr-beta[e]
Zr-MCM-41[e]
2-BuOH
8
2-BuOH
9
Zr(iPrO)4·iPrO
H
Al(iPrO)3
2-BuOH
10
11
2-BuOH
f
NaBH4
[a] Reaction conditions: 20 mg of estrone (0.08 mmol), alcohol (ca. 16 eq) and
Zr-catalyst (5 mg, ca. 18 mol% Zr), 393 K. [b] Conversion. Determined by GC.
Estradiols were the only products detected. [c] Turnover frequency. Moles of
estrone converted per mol of Zr and per hour of reaction. [d] α-E2:β-E2
diastereomeric ratio, calculated from the 1H NMR spectra of the reaction filtrates
(for the exact procedure used, see Supporting Information). [e] See Supporting
Information for a detailed description of these catalysts. [f] The procedure used
for estrone reduction with NaBH4 is described in the Experimental Section.
As we reported in our previous work, the use of MOF-808 as
catalyst for the MPV reduction of estrone produced a moderate
amount of α-E2, while most available biotransformation methods
and chemical routes produce almost exclusively the β-E2 isomer
(see entry 11 in Table 1 for the results obtained with NaBH4). This
is a very motivating result from the practical point given the current
interest in α-E2 for its anti-oxidative and anti-inflammatory
properties and as a potent ligand for a membrane-associated ER-
Figure 1. Reusability of MOF-808 for the MPV reduction of estrone with 2-BuOH at 393 K. (Left) Estrone conversion (white bars) and selectivity to α-E2 (-■-). (Right)
X-ray powder diffraction pattern of fresh MOF-808 (down) and MOF-808 recovered after 6 catalytic cycles (top).
3
This article is protected by copyright. All rights reserved.

10.1002/chem.202100967
Chemistry - A European Journal
FULL PAPER
Effect of the reaction temperature. The effect of reaction
temperature on the product distribution was assessed with both
iPrOH and 2-BuOH in the temperature range from 353 to 403 K.
Figure 2 shows the time-conversion plots obtained and the
corresponding α-E2: β-E2 d.r. measured at the end of the reaction
(white bars in the insets).
Figure 2. Time-conversion plots of MPV reduction of estrone over MOF-808 using isopropanol (left) or 2-butanol (right) in the
temperature range 353-403 K. In each plot, the inset shows the diastereomeric ratio of the reaction, d.r. (α-E2: β-E2).
Upon raising the reaction temperature from 353 K to 403 K, a
sharp and progressive increase of the reaction rate was observed
for both iPrOH and 2-BuOH. Thus, in the case of iPrOH, full
estrone conversion was attained after only 4 h of reaction at 403
K, while up to 48 h were needed to reach 83% conversion at 353
K. Thus, calculated TOFs increased from 0.78 h-1 to 9.11 h-1 in
the 353-403 K range. The same holds for 2-BuOH, for which TOF
increased from 0.33 to 3.78 h-1 from 353 to 403 K.
Kinetic analysis. Kinetic rate constants of α-E2 and β-E2
formation (hereafter kα and kβ) were calculated for both iPrOH and
2-BuOH in the 353-403 K temperature range, as detailed in the
Supporting Information. Apparent activation energies (Ea) for the
formation of α-E2 and β-E2 were calculated from the slopes of the
corresponding Arrhenius plots shown in Figure 3, while the
intercept with the y-axis was used to calculate the pre-exponential
factors, k0, which are correlated with entropic factors. To assess
the goodness of the fit, the values of Ea and k0 obtained for each
solvent were used to plot the theoretical evolution of the reaction
rate constant, k, as a function of the temperature according to the
formula k = k0 exp(−Ea/RT) and compared with the experimental
data (see Figure S6 in the Supporting Information).
Meanwhile, the diastereoselectivity of the reaction remained
almost unchanged within the whole temperature range and
depended only on the alcohol used (see insets in Figure 2). This
lack of kinetic or thermodynamic control is in agreement with the
reaction taking place inside the MOF cavities (where the Zr4+ are
located) and being affected mainly by cavity confinement effects
during the formation of the transition state (see below).
Figure 3. Arrhenius plots for the conversion of estrone to α-E2 and β-E2 in iPrOH (left) and 2-BuOH (right), with calculated apparent
activation energies (Ea) and pre-exponential factors (k0) for each compound.
4
This article is protected by copyright. All rights reserved.

10.1002/chem.202100967
Chemistry - A European Journal
FULL PAPER
The calculated activation energies are in general lower for
iPrOH than for 2-BuOH, reflecting the faster E1 conversion
observed for iPrOH (see Figure 2). For both iPrOH and 2-BuOH,
the formation of α-E2 and β-E2 have very similar apparent
activation energies, being only slightly lower for α- E2 than for β-
E2: 79.7 and 81.7 kJ/mol for iPrOH, and 83.2 and 84.9 kJ/mol for
2-BuOH. Meanwhile, the pre-exponential factors are in all cases
comparable, all of them of the same order of magnitude.
Given the small differences observed in both Ea and k0 values,
it is difficult at this point to determine what are the main factors
governing the higher α-E2 diastereoselectivity obtained with 2-
BuOH compared to iPrOH, or why the selectivity does not seem
to be affected by the reaction temperature. Moreover, if only the
differences in Ea between the two isomers (∆Ea) are considered,
one would expect a slightly higher selectivity towards α-E2 in
iPrOH than in 2-BuOH (since ∆Ea = 2 kJ/mol for iPrOH, and ∆Ea
= 1.7 kJ/mol for 2-BuOH). This obviously does not correspond to
the d.r. observed for this reaction. It is thus evident that entropic
contributions must play a relevant role in determining the
observed dependence of the reaction diastereoselectivity on the
alcohol used, as we will show below
While the Arrhenius equation is an empirical model, the
Eyring-Polanyi equation, derived from the Transition State Theory,
can be used to calculate directly the values of the entropy (∆S‡),
enthalpy (∆H‡), and Gibbs free energy (∆G‡) of activation for both
α- E2 and β- E2 and for the two alcohols used (see Figure S7 and
Table 2). Here, ∆G‡, ∆H‡, and ∆S‡ represent the changes in
energy upon forming an activated transition state complex
between estrone, the secondary alcohol, and the Zr4+ active site
of MOF-808. In this sense, we recall here that the generally
accepted mechanism for the MPV reduction of ketones over
Lewis acid catalysts assumes the simultaneous adsorption of the
alcohol and the ketone on the active site, involving a six-
membered cyclic transition state,[26] which in the case of estrone
should have the structure shown in Scheme 4.
Scheme 4. Proposed structure of the two possible transition
states for the MPV reduction of estrone leading to α-E2 and β-E2
over MOF-808, with either iPrOH or 2-BuOH.
As can be seen in Table 2, when the MPV reaction is carried
out in 2-BuOH at 393 K, the formation of the activated complex
leading to α-E2 is preferred over the other transition state by 4.1
kJ mol-1 (∆∆G‡ in bottom part in Table 2, see also Figure 4), while
this difference is only 1.4 kJ mol-1 in the case of iPrOH. Thus,
∆∆G‡ values in Table 2 directly reflect the higher d.r. obtained for
2-BuOH than for iPrOH: ca. 82% and 61% in the whole 353-403
K temperature range (see insets in Figure 2).
Table 2. Enthalpy, entropy and Gibbs free activation energy (∆H‡, ∆S‡ and ∆G‡)
for α-E2 and β-E2 over MOF-808 using iPrOH and 2-BuOH. The bottom part
reports the differential activation thermodynamic parameters, calculated as the
difference between the transition states leading to α-E2 and β-E2 (α−β) in terms
of Gibbs free energy (∆∆G‡), and breakdown into enthalpic (∆∆H‡) and entropic
contributions (-T∆∆S‡).
∆S‡
(J mol-1K-1)
-61.8
-T∆S‡ [a]
(kJ mol-1)
24.3
∆H‡
(kJ mol-1)
76.6
∆G‡ [a]
(kJ mol-1)
100.9
For a given solvent, we can also evaluate differential
activation parameters (∆∆G‡, ∆∆H‡ and ∆∆S‡), which correspond
to the difference between the transition states leading to α-E2 and
β-E2. These differential parameters are also reported in the
bottom part of Table 2 (α−β). To help interpret the meaning of
these parameters, Figure 4 shows a schematic representation
(not to scale) of the energy profile of the reaction with iPrOH and
2-BuOH, as well as simple snapshots representing the most
relevant structures formed in the course of the reaction.
iPrOH
α-E2
-60.5
-56.0
-61.9
∆∆S‡
(J mol-1 K-1)
-1.3
23.8
22.0
24.3
-T∆∆S‡ [a]
(kJ mol-1)
+0.5
78.5
80.0
81.8
∆∆H‡
(kJ mol-1)
-1.9
102.3
102.0
106.1
∆∆G‡ [a]
(kJ mol-1)
-1.4
β-E2
α-E2
β-E2
2-BuOH
iPrOH
α−β
2-BuOH
+5.9
-2.3
-1.8
-4.1
[a] Calculated at 393 K.
5
This article is protected by copyright. All rights reserved.

10.1002/chem.202100967
Chemistry - A European Journal
FULL PAPER
Figure 4. Schematic energy diagram (not to scale) showing the differences in energies between the two activated complexes leading to α-E2 and β-E2 in iPrOH
(left) and 2-BuOH (right). The bottom part show simple snapshots representing the most relevant structures formed in the course of the reaction.
For the reaction in 2-BuOH, the entropy loss upon the
formation of the transition state is lower for α-E2 than for β-E2
(-56.0 vs -61.9 J mol-1 K-1). Therefore, both the differential
enthalpic (∆∆H‡ = -1.8 kJ/mol) and entropic contributions (-T∆∆S‡
= -2.3 kJ/mol) have the same sign, and both contribute to lowering
the energy of the transition state of α-E2 compared to β-E2. On
the contrary, for the reaction in iPrOH, the differential enthalpic
and entropic terms have opposite signs (-1.9 and +0.5 kJ/mol,
Diastereoselective reduction of 5α-androstan-3β-ol,17-one
(epiandrosterone)
Given the excellent diastereoselectivity attained with MOF-808 for
the reduction of estrone to 17α-estradiol, we wanted to extend the
study to the reduction of other ketosteroids. In this case, we
considered the reduction of epiandrosterone (hereafter EPIA),
which also bears a keto group in position 17. Similar to estrone,
EPIA reduction can yield two diastereoisomers, 5α-androstan-
3β,17α-diol (α-EPIAdiol) and 5α-androstan-3β,17β-diol (β-
EPIAdiol), depending on the relative conformation of the 17-OH
group with respect to the 18-CH3 group (Scheme 5):
respectively), partially counteracting each other. As
a
consequence, the energy difference between both transition
states is much lower in iPrOH.
Taken the above data together, our results thus suggest that
the space available inside the MOF cavities can accommodate
better the transition state TS-α than TS-β, and this difference is
higher with 2-BuOH than with iPrOH. This is not surprising if we
consider that 2-BuOH (or iPrOH) directly participate in the
formation of the activated complex (see Scheme 4).
Therefore, in the case of 2-BuOH, the important contribution
from the entropic term on the total energy difference between the
two possible transition states strongly indicates that confinement
effects and steric hindrance inside the MOF cavity are the main
reasons determining the preferred reaction pathway and the
observed diastereoselectivity. Accordingly, the selectivity to α-E2
is much lower for the reaction with iPrOH, for which the entropic
term partially neutralizes the enthalpic contribution, and the
energy difference is therefore lower.
Scheme 5. Reduction of epiandrosterone (EPIA) can yield a mixture of
androstandiols (α-EPIAdiol and β-EPIAdiol)
Analogous to the case of estrone reduction, the presence of
the CH3 group in position 18 introduces a strong steric hindrance
to the hydride addition to the carbonyl from the upper face, so β-
EPIAdiol is usually the only product obtained in most synthetic
routes. Thus, the preparation of the elusive α-EPIAdiol compound
is a challenging synthetic process. Nevertheless, α-EPIAdiol can
6
This article is protected by copyright. All rights reserved.

10.1002/chem.202100967
Chemistry - A European Journal
FULL PAPER
be an interesting target for the treatment of cancer, due to anti-
androgenic properties, and in the treatment of osteoporosis,[27,28]
so new selective synthetic routes are also sought for this
compound.
To our delight, MPV reduction of EPIA using MOF-808 as
catalyst produces α-EPIAdiol as the main product with excellent
diastereoselectivities, depending on the secondary alcohol used:
77% d.r. and 84% d.r. in the case of iPrOH and 2-BuOH,
respectively, as shown in Figure 5. In both cases, androstandiols
were the only products observed, and the catalyst was found to
be stable and reusable.
substrate to assess both chemo- and regioselectivity of the MPV
reduction using MOF-808 catalyst.
In our hands, upon reduction of A4 over MOF-808 with either
iPrOH or 2-BuOH, we did not observe the formation of products
coming from the reduction of the C=C bond (100%
chemoselectivity) or the mono-reduction of the carbonyl group in
position 3 (see Supporting Information for more details on product
1
identification by H and 13C-NMR). Therefore, the only products
detected were testosterone (T) and epitestosterone (E) from the
reduction of the 17 keto group, and a mixture of diol isomers
coming from the reduction of both keto groups, as shown in
Scheme 6:
Scheme 6. MPV reduction of androstenedione (A4) over MOF-808 yields
testosterone (T), epitestosterone (ET) and a mixture of diol isomers with
different local configurations of the OH groups at positions 3 and 17.
Figure 6 shows the time-conversion plots and product
distribution obtained with 2-BuOH at 353 K, while iPrOH afforded
very similar results (see Figure S8 in the Supporting Information).
Figure 5. MPV reduction of epiandrosterone (EPIA) to androstandiols using
iPrOH and 2-BuOH. The obtained diastereoselectivity (d.r.) to α-EPIAdiol is
indicated. d.r. was calculated by 1H NMR of the filtrate at the end of the reaction
(see Supporting Information). Reaction conditions: 23 mg of EPIA (0.08 mmol),
alcohol (ca. 16 eq) and Zr-catalyst (5 mg, ca. 18 mol% Zr), 393 K.
Together with estrone, the results obtained for the MPV
reduction of EPIA confirm the general applicability of MOF-808 as
a stable and reusable catalyst for the diastereoselective reduction
of 17-ketosteroids to the corresponding 17α-hydroxysteroids.
Thus, MPV reduction of ketosteroids over MOF-808 represents a
very attractive and reliable one-step catalytic alternative for the
existing stoichiometric reactions using hydride transfer
compounds. As mentioned in the introduction, most existing
chemical methods usually yield 17β-OH compounds, so
preparation of 17α-OH isomers require several synthetic steps
and they use toxic and/or expensive reagents, leading to very low
final yields of the target compound.
Regio- and diastereoselective reduction of androstenedione
To increase further the scope of MOF-808 as a catalyst for the
MPV reduction of ketosteroids, it would be highly desirable to
develop a catalyst not only diastereoselective but also chemo-
and regioselective. It is well-known that MPV is, in general, a
chemoselective process, which only transforms carbonyl groups
whilst leaving other easily reducible functional groups intact, as in
the case of unsaturated ketones.[29] However, a more demanding
situation is found when the substrate contains more than one
carbonyl group that, in principle, can be reduced, leading to a
mixture of mono- and polyhydroxylated compounds. In this case,
the regioselectivity of the reduction process may become an issue.
∆4-androstene-3,7-dione (androstenedione, hereafter A4)
contains a C=C bond in position 4 and two keto groups in
positions 3 and 17, so we considered it as an appropriate
Figure 6. (■) A4 conversion; (○) yield of mono-hydroxylated (T+ET); and (▲)
yield of dihydroxylated products obtained over MOF-808 with 2-BuOH at 353 K.
Reaction conditions: 10 mg of A4 (0.035 mmol), 1 mL 2-BuOH (ca. 30 eq) and
10 mg MOF-808.
According to the time evolution of products shown in Figure 6,
A4 reduction at the 17 position occurs in the first place, yielding a
mixture of mono-hydroxylated T and ET as primary products,
followed afterwards by the reduction of the keto group also in
position 3, to form the corresponding 3,17-diols, at longer reaction
times. Thus, the amount of 17-mono-hydroxylated products
increases during the first 8 hours of reaction, reaching a maximum
yield of 58%, and then starts to decrease gradually as more diols
are formed.
7
This article is protected by copyright. All rights reserved.

10.1002/chem.202100967
Chemistry - A European Journal
FULL PAPER
Identification of the products using 13C and 1H NMR
spectroscopy (see Supporting Information) revealed that the
amount of T formed was very low, close to the detection limits of
the technique, and in any case, less than 5% referred to ET.
Therefore, as in the case of estrone and epiandrosterone
discussed above, MOF-808 yielded the corresponding 17α-OH
inhibitory effects of 5alpha-reductase,[30] and it has agonistic
effects
on
a
Prostate-Specific
G-Protein
Receptor
(PSGR/OR51E2).[31]
In an attempt to improve the selectivity to mono-hydroxylated
products (T and ET) and minimize the formation of diols, the
reaction temperature was varied within the 333-393 K range.
Figure 7 shows the A4 conversion curves and corresponding
compound
(viz,
epitestosterone)
with
excellent
diastereoselectivity over the 17β-compound, higher than 95%.
This result is also interesting from the practical point of view, since
ET has applications in the treatment of prostate cancer, due to its
selectivity-conversion plots obtained as a function of the
temperature. The complete kinetic data is given in Table S1 in the
Supporting Information.
Figure 7. (Left) A4 conversion over MOF-808 in 2-BuOH at the indicated temperature. (Right) Selectivity to monohydroxylated products,
ET+T, as a function of the A4 conversion and temperature.
As it is seen in Figure 7, the maximum selectivity to
Herein we have shown that MOF-808 is a very interesting
monohydroxylated products is around 85% at 10-20% conversion, heterogeneous catalyst for the selective MPV reduction of
but this value decreases gradually at higher levels of conversion,
dropping faster as the reaction temperature increases. Therefore,
increasing the reaction temperature has a clear detrimental effect
on the selectivity to mono-hydroxylated products. Nevertheless,
our data show that if the reaction temperature is kept at 333 K, it
is possible to attain a level of A4 conversion as high as 70-80%
ketosteroids to the corresponding hydroxysteroids. On the one
hand, the particular structure of the Zr6 clusters in MOF-808 (after
activation and removal of the capping formate anions) provides
coordination vacancies for substrate binding and transformation.
On the other hand, the wide pore structure of MOF-808 compared
to the steroid dimensions ensure a good diffusion of the reaction
substrates and products to and from the Zr4+ active sites.
Consequently, MOF-808 is a highly active catalyst for the MPV
reduction, requiring the simultaneous coordination of the carbonyl
substrate and the secondary alcohol used as a hydride source.
Interestingly, we have shown that reduction of 17-ketosteroids
produces the corresponding 17α-OH compound with very high
diastereoselectivity, whereas most existing biotransformation
methods and chemical routes yield the other isomer, 17β-OH, as
the main product. Therefore, MOF-808 provides a very attractive
catalytic route for the preparation of these challenging
compounds, with several improvements compared to alternative
bio- and chemical processes. We have demonstrated the
excellent diastereoselectivity of MOF-808 for the synthesis of
three different 17α-hydroxysteroids; namely, 17α-estradiol,
androstan-3β,17α-diol, and epitestosterone.
Temperature-dependent kinetic analysis has been used to
calculate reaction rate constants, apparent activation energies
and thermodynamic parameters of estrone reduction using either
iPrOH or 2-BuOH. A close inspection of the energy differences
between the two transition states leading to α-E2 and β-E2
revealed a major contribution from the entropic term in the case
of 2-BuOH, while in the case of iPrOH the entropic contribution
partially neutralizes the enthalpic gain, thus lowering the energy
differences between both transitions states. These findings
after 30-35 hours, while maintaining
a
selectivity to
monohydroxylated compounds above 80% (see the arrow in
Figure 7) and a diastereoselectivity >95% to ET, which is
considered an excellent result.
Overall, our study has revealed that MOF-808 affords a very
good selectivity for the reduction of A4 to ET, since: i) the C=C
bond in position 4 remains intact under the reaction conditions
used (100% chemoselectivity); ii) no traces of 3-hydroxy-17-
ketosteroids have been detected, so the 17-carbonyl group is
selectively reduced before the 3-carbonyl (100% regioselectivity);
iii) reduction of the 17-keto group produces epitestosterone
almost exclusively (>95%); i.e., the 17α-OH isomer is selectively
formed, as in the case of estrone and epiandrosterone (excellent
diastereoselectivity); and iv) despite the eventual reduction of
both carbonyl groups taking place at long reaction times, it is
possible to minimize the formation of diols by proper selection of
the reaction conditions, allowing very high A4 conversions (70-
80%) while keeping very good selectivity (>80% to
monohydroxylated compounds).
Conclusion
8
This article is protected by copyright. All rights reserved.

10.1002/chem.202100967
Chemistry - A European Journal
FULL PAPER
strongly suggest that the diastereoselectivity to α-E2 originates
from the confinement of the reaction inside the MOF cavities, in
which the Zr4+ active sites are located. This cavity confinement
inside the MOF drives the reaction preferentially through the α-E2
transition state in both iPrOH and 2-BuOH, but this preference is
higher with 2-BuOH, in line with the direct participation of the
secondary alcohol in the six-membered cyclic transition state.
Finally, MPV reduction of androstenedione, containing a C=C
bond and two keto groups, has been used to evidence the
excellent chemo-, regio- and diastereoselectivity of the reaction
catalyzed by MOF-808.
Acknowledgements
Financial support by the Spanish Government is acknowledged
through projects MAT2017-82288-C2-1-P and the Severo Ochoa
program (SEV-2016-0683). The Microscopy Service of the
Universitat Politècnica de València is gratefully acknowledged for
the electron microscopy images.
Keywords: Steroids • metal-organic frameworks • MOF-808 •
Meerwein-Ponndorf-Verley • heterogeneous catalysis
[1]
[2]
G. P. Moss, Pure Appl. Chem. 1989, 61, 1783–1822.
F. Buttgereit, G. R. Burmester, B. J. Lipworth, Lancet 2005, 365,
801–803.
Experimental Section
[3]
[4]
M. V. Donova, O. V. Egorova, V. M. Nikolayeva, Process Biochem.
2005, 40, 2253–2262.
Catalyst preparation.
M. A. Faramarzi, N. Zolfaghary, M. Tabatabaei Yazdi, S. Adrangi, H.
Rastegar, M. Amini, M. Badiee, J. Chem. Technol. Biotechnol. 2009,
84, 1021–1025.
MOF-808. MOF-808 was prepared with slight modifications from an earlier
reported procedure by Furukawa et. al.[13] Briefly, a solution was prepared
with 242,5 mg of ZrOCl2·8H20 (0,75 mmol), 105 mg of trimesic acid (0,5
mmol) and 22,5 mL of a DMF/HCO2H 1:1 (v/v) mixture. The solution was
transferred into a Teflon lined autoclave and heated inside an oven at
130 °C for 48 h. After cooling down to room temperature, the material was
recovered by centrifugation and washed for 3 days with DMF (changing
the solvent 2 times per day) and for another 3 days with EtOH (changing
the solvent 2 times per day). After removing the solvent by centrifugation,
the solid was dried in air. X-ray diffraction (PhillipsX’Pert, Cu Kα radiation)
was used to confirm the expected structure type and high crystallinity of
the material. According to N2-adsorption experiments, calculated specific
surface area (SBET) and pore volume were 1345 m2g-1 and 0.60 cm3g-1,
respectively. A summary of the most relevant characterization data of this
material is shown in Figure S1 in the Supporting Information.
[5]
[6]
W. H. Liu, C. W. Kuo, K. L. Wu, C. Y. Lee, W. Y. Hsu, J. Ind.
Microbiol. 1994, 13, 167–171.
O. Schaaf, K. Dettner, J. Steroid Biochem. Mol. Biol. 1998, 67, 451–
465.
[7]
[8]
[9]
F. Balssa, M. Fischer, Y. Bonnaire, Steroids 2014, 86, 1–4.
J. A. Dodge, C. W. Lugar III, Bioorg. Med. Chem. Lett. 1996, 6, 1–2.
T. Nambara, M. Numazawa, H. Takahashi, Chem. Pharm. Bull.
(Tokyo). 1969, 17, 1725–1729.
[10]
X. X. Shi, C. L. Shen, J. Z. Yao, L. D. Nie, N. Quan, Tetrahedron
Asymmetry 2010, 21, 277–284.
[11]
[12]
P. P. Guo, K. Ding, Tetrahedron Lett. 2015, 56, 4096–4100.
G. Göndös, J. C. Orr, J. Chem. Soc. Chem. Commun. 1982, 1238–
1239.
Other catalysts: The Supporting Information contains details of the
preparation of other non-commercial catalysts used in this work for
comparison purposes.
[13]
[14]
H. Furukawa, F. Gandara, Y. B. Zhang, J. Jiang, W. L. Queen, M. R.
Hudson, O. M. Yaghi, J. Am. Chem. Soc. 2014, 136, 4369–4381.
A. H. Valekar, K.-H. Cho, S. K. Chitale, D.-Y. Hong, G.-Y. Cha, U.-
H. Lee, D. W. Hwang, C. Serre, J.-S. Chang, Y. K. Hwang, Green
Chem. 2016, 18, 4542–4552.
Catalytic experiments
MPV reduction of ketosteroids. Particular reaction conditions for each
substrate are indicated in the text. To illustrate the typical procedure used,
reduction of estrone over MOF-808 with iPrOH will be described. Estrone
(20 mg, 0.08 mmols) was dissolved in iPrOH (1 mL, ca. 16 eq.) and
introduced inside a screw cap glass reactor containing MOF-808 (5 mg,
ca. 18 mol% Zr referred to estrone) and kept at the selected reaction
temperature under magnetic stirring (stirring rate was kept > 600 rpm to
ensure optimal mixing and to avoid diffusional limitations). Time-evolution
of products was analyzed by GC (Agilent 7890 A, equipped with a FID
detector and a DB5 30 m x 0,25 mm x 0,25 µm column) on sample aliquots
taken at fixed time intervals. Product identification was done by GC-MS
and by comparison with pure standards. Peak separation of the different
diastereoisomers in the resulting chromatograms was good enough to
extract the concentration of the individual species. More accurate
diastereomeric ratios were obtained at the end of the reaction from the
corresponding 1H NMR spectra, as detailed in the Supporting Information.
[15]
[16]
[17]
[18]
[19]
E. Plessers, G. Fu, C. Y. X. Tan, D. E. de Vos, M. B. J. Roeffaers,
Catalysts 2016, 6, 104.
H.-H. Mautschke, F. Drache, I. Senkovska, S. Kaskel, F. X. Llabrés i
Xamena, Catal. Sci. Technol. 2018, 8, 3610–3616.
J. H. Cavka, S. Jakobsen, U. Olsbye, N. Guillou, C. Lamberti, S.
Bordiga, K. P. Lillerud, J. Am. Chem. Soc. 2008, 130, 13850–13851.
F. G. Cirujano, A. Corma, F. X. Llabrés i Xamena, Chem. Eng. Sci.
2015, 124, 52–60.
C. D. Toran-Allerand, A. A. Tinnikov, R. J. Singh, I. S. Nethrapalli,
Endocrinology 2005, 146, 3843–3850.
[20]
[21]
S. P. Kaur, S. Bansal, K. Chopra, Steroids 2015, 96, 7–15.
C. Behl, T. Skutella, F. Lezoualc’h, A. Post, M. Widmann, C. J.
Newton, F. Holsboer, Mol. Pharmacol. 1997, 51, 535–541.
J. A. Dykens, W. H. Moos, N. Howell, Ann. N. Y. Acad. Sci. 2005,
1052, 116–135.
Estrone reduction with NaBH4. A solution was prepared by dissolving
estrone (40 mg, 0.15 mmols) in ethanol (5 mL). A second ethanolic solution
(1 mL) containing NaBH4 (18.9 mg) was added to the first one at 353 K
under stirring and kept at this temperature for 15 min. Then, the
temperature was raised to reflux, and the heating was stopped just after
the boiling started, and let it stand for 5 min. To stop the reaction, 10 mL
of H2O was added and the mixture was cooled in a water-ice bath for 15
min, whereupon a white precipitate of the product formed. Finally, the
product was recovered by filtration and washed with cold water and
analyzed by GC as described above.
[22]
[23]
[24]
[25]
R. S. Santos, L. A. De Fatima, A. P. Frank, E. M. Carneiro, D. J.
Clegg, Biol. Sex Differ. 2017, 8, 1–13.
G.-D. Han, Z.-Y. Lin, C.-P. Yan, Zhongguo Yiyao Gongye Zazhi
1994, 25, 490–491.
T. Ohta, H. Zhang, Y. Torihara, T. Hida, I. Furukawa, Steroids 1998,
63, 630–632.
9
This article is protected by copyright. All rights reserved.

10.1002/chem.202100967
Chemistry - A European Journal
FULL PAPER
[26]
[27]
[28]
C. F. de Graauw, J. A. Peters, H. Van Bekkum, J. Huskens,
Synthesis (Stuttg). 1994, 10, 1007–1017.
P. Wang, Y. Wen, G. Z. Han, P. K. Sidhu, B. T. Zhu, Br. J.
Pharmacol. 2009, 158, 1796–1807.
T. F. H. Bovee, M. Thevis, A. R. M. Hamers, A. A. C. M.
Peijnenburg, M. W. F. Nielen, W. G. E. J. Schoonen, J. Steroid
Biochem. Mol. Biol. 2010, 118, 85–92.
[29]
M. De Bruyn, M. Limbourg, J. Denayer, G. V. Baron, V. Parvulescu,
P. J. Grobet, D. E. De Vos, P. A. Jacobs, Appl. Catal. A Gen. 2003,
254, 189–201.
[30]
[31]
B. A. Nuck, A. W. Lucky, J. Invest. Dermatol. 1987, 89, 209–211.
T. Abaffy, H. Matsunami, Modulators of Prostate-Specific G-Protein
Receptor (PSGR/OR51E2) and Methods of Using Same, 2018,
US2018116992.
10
This article is protected by copyright. All rights reserved.

10.1002/chem.202100967
Chemistry - A European Journal
FULL PAPER
Entry for the Table of Contents
Zr-containing MOF-808 is a stable and reusable catalysts for the chemo-, regio- and diastereoselective reduction of 17-ketosteroids to
elusive 17α-hydroxysteroids compounds, as we shoed for the synthesis of 17α-estradiol, androstan-3β,17α-diol, and epitestosterone.
The process is carried out in one single step and without using toxic and/or expensive reagents. A temperature-dependent kinetic
analysis of estrone reduction revealed that the diastereoselectivity to 17α-estradiol originates from the confinement of the reaction
inside the MOF cavities, in which the Zr4+ active sites are located.
11
This article is protected by copyright. All rights reserved.