Non-hazardous Baeyer-Villiger oxidation of levulinic acid derivatives: Alternative renewable access to 3-hydroxypropionates

Article abstract of DOI:10.1039/c4cc08734h

Baeyer-Villiger monooxygenases catalyze the energetically challenging oxidation of levulinates (4-oxopentanoates) to 3-hydroxypropionic acid (3-HPA) derivates under ambient conditions, replacing propellant-grade H₂O₂ with aerial oxygen as the oxidant. This reaction enables a new pathway to a platform for chemical 3-HPA, an important intermediate in the non-petrol based production of a variety of bulk chemicals (acrylates, malonates, 1,3-propanediol).

Full text of DOI:10.1039/c4cc08734h

ChemComm
View Article Online
View Journal
COMMUNICATION
Non-hazardous Baeyer–Villiger oxidation of
levulinic acid derivatives: alternative renewable
Cite this:DOI: 10.1039/c4cc08734h
access to 3-hydroxypropionates†
Received 7th November 2014,
Accepted 5th January 2015
Michael J. Fink and Marko D. Mihovilovic*
DOI: 10.1039/c4cc08734h
www.rsc.org/chemcomm
Baeyer–Villiger monooxygenases catalyze the energetically challenging
oxidation of levulinates (4-oxopentanoates) to 3-hydroxypropionic acid
(3-HPA) derivates under ambient conditions, replacing propellant-grade
H2O2 with aerial oxygen as the oxidant. This reaction enables a new
pathway to a platform for chemical 3-HPA, an important intermediate
in the non-petrol based production of a variety of bulk chemicals
(acrylates, malonates, 1,3-propanediol).
Fig. 1 BVOx of alkyl levulinates 1–5. a m-Chloroperbenzoic acid or t-butyl
hydroperoxide in DCM or toluene; peracetic acid or trifluoroperacetic acid
generated in situ from 50% H2O2 and the corresponding carboxylic acid as
Levulinic acid (LevOH) is generated as a by-product from carbo-
hydrate dehydration in a significant percentage, and yearly tonnage solvent; for 5: 90% H2O2, trifluoroacetic anhydride, DCM, 63% (ref. 5).
is therefore available at a competitively low price.1 Baeyer–Villiger
oxidation (BVOx) of levulinates possibly leads to succinates upon
unfavored methyl migration, or to 3-acetoxypropionates, which
readily hydrolyze to give esters of 3-HPA. This compound is a
versatile C3 intermediate in the production of acrylates, malonates,
1,3-propanediol and other similar bulk chemicals.2 It is currently
being used as a crosslinking agent for polymer coatings and metal
lubricants, and as an antistatic agent for textiles. More importantly,
it was mentioned in third place in the current Top Value Added
Chemicals from Biomass list of the US Department of Energy as a
sugar-derived building block.3 Bio-based access from levulinates to
this key intermediate may thus create an alternative to direct
fermentative approaches towards the formation of 3-HPA (from
glucose or glycerol), while valorizing LevOH waste; so far, no
fermentative process has achieved a level of economic viability
on an industrial scale.4
In this work we present the biocatalytic BVOx of alkyl levulinates
(LevOR, R = alkyl) to 3-acetoxypropionates and 3-hydroxypropionates
on the gram scale under ambient conditions using aerial oxygen as
the terminal oxidant. In the preceding literature the uncatalyzed
BVOx of benzyl levulinate was the only example available, carried
out using 90% H2O2 and trifluoroacetic anhydride in refluxing DCM
and yielding the expected acetoxy product 5a in 63% yield (Fig. 1).5
Since H2O2 at this concentration is mainly used as a rocket
propellant6 and the reaction was carried out on the 0.44 mol scale,
we chose not to reproduce this experiment for safety reasons.
The first attempts to convert the initial test substrates
LevOMe, LevOEt and LevOnBu (1–3) using more common and
comparably safe peroxide sources and activating agents were
poorly successful (Fig. 1): only trifluoroperacetic acid at room
temperature gave 24–29% conversion to a mixture of defined
products (including 1a–3a) after 48 h. All other reagents and
conditions were either unreactive or led to complete decomposi-
tion of the starting materials (see ESI†).
LevOH is known to form a-angelica lactone via intramolecular
enol lactonization under strongly acidic and dehydrating condi-
tions.7 The intermediate enol tautomer would be impervious to a
nucleophilic peroxide attack, but no evidence for the enol form
and/or intramolecular transesterification was found in comparative
NMR analyses of CDCl3 and CF3CO2D solutions of 1–3 (data not
shown). These findings led us to believe that BVOx of levulinates
are energetically demanding due to the stereoelectronic require-
8
´
ments for a successful rearrangement of the Criegee intermediate,
which obviously cannot be overcome by increased redox potential,
nucleophile quality or concentration-kinetic effects (activation
barriers for unreactive conformers have been shown to be higher
by as much as 200 kJ molÀ1 via DFT calculations).9
Baeyer–Villiger monooxygenases (BVMOs), a subclass of flavin-
and NADPH-dependent monooxygenases, are well-studied bio-
catalysts for BVOx.10 Their ability to induce chirality in this
Institute of Applied Synthetic Chemistry, Vienna University of Technology,
Getreidemarkt 9/163, 1060 Vienna, Austria.
E-mail: marko.mihovilovic@tuwien.ac.at
† Electronic supplementary information (ESI) available: Experimental details,
copies of NMR spectra. See DOI: 10.1039/c4cc08734h
This journal is ©The Royal Society of Chemistry 2015
Chem. Commun.

View Article Online
Communication
ChemComm
Fig. 2 Analytical scale screening of wild-type BVMOs for the oxidation of levulinates 1–3; relative conversion as determined by GC-MS(EI); reaction
conditions: BVMO in recombinant E. coli BL21(DE3), air, LB medium, 4 mM substrate, 0.5% v/v 1,4-dioxane, 1 mL reaction volume, 24 1C, 24 h.
reaction class at room temperature is so far unsurpassed by any
other catalytic system, which in turn highlights their excellent
ability to (specifically) lower activation energy. Analytical scale
whole-cell enzyme screening was performed on esters 1–3 with
a selection of 13 BVMOs: seven cyclohexanone monooxygenase-
type (CHMO), two cyclopentanone monooxygenase-type (CPMO),
two linear ketone-converting and two large ring-converting
BVMOs. This demonstrated the excellent stereoelectronic control
of these catalysts: the substrates were partly converted to the
acetoxy products 1a–3a using aerial oxygen under ambient con-
ditions by nine biocatalysts (Fig. 2). Interestingly, the best results
Fig. 3 Concentration screening of levulinates 1–5 using whole-cell BVMO
were achieved with CPMOComa (ref. 11) and cyclododecanone
catalysts; at all concentrations no visible formation of a second liquid phase
was observed; relative conversion determined by GC-FID; LevOMe 1 not done
with CDMORhodo; x-axis ticks indicate tested initial substrate concentrations.
12
monooxygenase CDMORhodo
.
On the other hand, both linear
ketone-converting BVMOs ( p-hydroxyacetophenone monooxygenase
HAPMOP.fl. (ref. 13) and phenylacetone monooxygenase
14
PAMOThermo
)
failed to convert this open-chain ketone (3a was an aqueous environment at sub-physiological pH and room tem-
detected in trace amounts using HAPMOP.fl.). These findings fit perature. Nevertheless, both product compounds could easily be
into the general substrate profiles of the two best-performing isolated from the medium by ethereal solvent extraction. In any
wild-type catalysts: CPMOComa is known to preferentially stabi- case, no regioisomeric oxidation products (methyl alkyl succinates)
´
lize only one conformer in the Criegee intermediate, resulting in were detected.
regioisomerically homogeneous products.15 CDMORhodo was
Next, we tested a CAST library of CPMO variants in order to
found to be an excellent catalyst for the oxidation of linear further improve the productivity of this reaction. This library
b-amino ketones, among other substrate classes.16 In addition was already previously generated and described with the intent
to the enzyme-specific interpretation, a clear positive correla- of improving the selectivity and/or extend the substrate accept-
tion between observed reaction rate and ester chain length was ability of wild-type CPMO.17 Four active site positions had been
observed. Esters of longer chain alcohols are converted faster targeted (F156, G157, G449, F450), modulating available space
and are tolerated at higher substrate concentrations (Fig. 3). and hydrogen bonding properties specifically without rando-
So far, we have not been able to determine to what extent this mization. Four double mutants were also included. In order to
effect is based on enhanced lipophilicity and thus faster trans- screen for enhanced activity, further tests were performed at 8 mM
membrane transport of substrates in whole-cell biocatalysis.
substrate loading (Fig. S3, ESI†). Two mutants showed significantly
The product structure and integrity of the analytical results were increased activity with at least one of the substrates: CPMOF156L
confirmed with preparative biotransformations in shake flasks converted esters 1–3 almost completely within 24 h, and F450I
using CPMOComa (esters 1–3) and CDMORhodo (esters 2 and 3) on exchange improved the reaction rate for LevOnBu. We hypothesize
the 100 mg scale. We obtained the expected 3-acetoxypropionates that these two amino acids are close to the ester residue, since
1a–3a, but missing pH regulation in the simple vessels partly led to reduced steric bulk results in higher activity, and substitution of
the cleavage of the acetate moiety during the reaction and/or the native phenylalanine to non-aromatic amino acids significantly
workup; only methyl ester 1a remained intact and was obtained decreased conversion of LevOBn 5, suggesting p–p interactions in
in 80% yield. This information proved to be valuable: hydrolysis of the wild-type (Fig. S1 and S2, ESI†). Moreover, the previously
the BVOx products to 3-hydroxypropionates proceeds smoothly in observed activity trend seemed to be reversed or not present in
Chem. Commun.
This journal is ©The Royal Society of Chemistry 2015

View Article Online
ChemComm
Communication
the mutant F156L. Within this mutant study, we also tested a
18
CPMO-templated quadruple mutant of PAMOThermo
,
with
positive results. In contrast to its parent enzyme, PAMO15-F5
was active on esters 2 and 3, further substantiating this strategy
in enzyme engineering (Fig. S3, ESI†).
Eventually, CPMOF156L was chosen as the best candidate for
biotransformation up-scaling and process development. LevOEt
2 was used as the model substrate based on the compromise
between favorable kinetics and atom economy. Moreover, ethyl
3-HPA was explicitly stated as a desired commodity chemical
target.3 Oxidation of 2 using whole-cells under monitored condi-
tions in a bioreactor on a 2.6 g scale (1.8 L reaction volume, 10 mM
initial substrate titer, 20 h reaction time) gave 2a in 66% yield and
high purity after simple extraction of the centrifuged culture
medium. Uncontrolled hydrolysis of the acetate was avoided by
maintaining a neutral pH during the reaction and workup. This
results in a space-time yield (STY) of 1.9 g LÀ1 dÀ1 by conversion,
or 1.3 g LÀ1 dÀ1 based on the isolated product.
Fig. 5 Comparison of concentration effects of 2 on CCE and whole-cells
of E. coli BL21(DE3) expressing CPMOF156L; top: 1% w/v CCE, neat substrate
addition, 2% v/v MeOH co-solvent for experiments 420 mM substrate;
bottom: growing cells (24 1C), induction and substrate addition after a 2 h
incubation time (37 1C), 0.5% v/v 1,4-dioxane as the co-solvent for all
experiments; GC-FID analysis.
We further investigated the limits of the whole-cell reaction
mode and observed the significant toxicity at 410 mM of 2
(1.44 g LÀ1, Fig. 4). This represented a serious obstacle for
improvements along this line. Thus, using a solid polystyrene
substrate and product reservoir (Lewatit VP OC 1163),19 we were
able to maintain a non-toxic substrate titer and increase the STY to
2.4 g LÀ1 dÀ1 in small scale experiments (100 mL; data not shown).
In order to circumvent the unexpectedly strong toxicity of 2 to the
recombinant host organism, we opted for crude cell extract (CCE)
reactions at the cost of xenobiotic enzyme denaturation and its
intrinsic decay. For economic viability, in situ NADPH recycling was
facilitated by a secondary enzymatic system (glucose-6-phosphate
dehydrogenase). In comparison with whole-cell reactions, the pro-
ductivity of CCE biotransformations was increased up to a concen-
tration of 20 mM LevOEt (Fig. 5). Although no substrate inhibition
and/or denaturation of the enzyme was observed up to a titer of
150 mM 2 (approx. 22 g LÀ1), CPMOF156L seems to be strongly
inhibited by the reaction product 2a at 420 mM available
aqueous concentration. This was cross-validated in a competition
experiment: a CCE solution pre-saturated with 20 mM of 2a could
still fully oxidize 1 mM LevOEt (data not shown).
Nevertheless, this reaction setup leads to a calculated average
productivity of 5 mM hÀ1 (STY approx. 17 g LÀ1 À1), as long as
d
the resulting BVOx product can continuously be removed from
the aqueous phase. The compatibility of such a substrate feed
and product removal system (as described above with whole cells)
was already reported for other redox biocatalysts.20 We are
currently investigating the feasibility of this concept with BVMOs.
In summary, we have demonstrated another instance of the
potential of BVMOs as catalysts in achiral transformations: a
propellant-grade stoichiometric oxidant in a corrosive organic
solvent mixture could be replaced with aerial oxygen in water
under ambient conditions. Levulinates, by-products from biomass
degradation, could be converted to derivatives of 3-HPA, a key
building block for many petrol-dependent bulk chemicals (acrylates,
malonates, 1,3-propanediol) in a safe way. Using generically
designed variants of CPMOComa it was possible to convert the
model substrate LevOEt on the gram scale. We are confident
that tailored enzyme engineering is capable of further boosting
the productivity of this transformation beyond the lab scale.
We thank Dr Martin Schu¨rmann (DSM) and Dr Robert
DiCosimo (DuPont) for suggestions and scientific advice, and
Prof. Margaret M. Kayser (University of New Brunswick) and
Prof. Manfred T. Reetz (Max Planck Institute for Coal Research,
Mu¨lheim an der Ruhr) for the kind donation of the CPMO
variants. PAMO15-F5 was provided by Prof. Marco W. Fraaije
within the Oxygreen consortium (EU-FP7 grant # 212281). M.J.F.
was also funded under this program.
Notes and references
1 D. M. Alonso, J. Q. Bond and J. A. Dumesic, Green Chem., 2010, 12,
1493–1513; D. W. Rackemann and W. O. S. Doherty, Biofuels,
Bioprod. Biorefin., 2011, 5, 198–214; S. G. Wettstein, D. M. Alonso,
E. I. Gurbuz and J. A. Dumesic, Curr. Opin. Chem. Eng., 2012, 1,
218–224.
Fig. 4 Growth curves of recombinant E. coli BL21(DE3) expressing CPMOF156L
at various conc. of LevOEt 2; cultures were inoculated to OD600 = 0.05 from an
overnight culture and grown at 37 1C; after 2 h substrate was added and
cultivation was continued at 24 1C (expression temp.).
This journal is ©The Royal Society of Chemistry 2015
Chem. Commun.

View Article Online
Communication
ChemComm
2 C. Della Pina, E. Falletta and M. Rossi, Green Chem., 2011, 13, 13 N. M. Kamerbeek, M. J. H. Moonen, J. G. M. van der Ven, W. J. H.
1624–1632.
van Berkel, M. W. Fraaije and D. B. Janssen, Eur. J. Biochem., 2001,
268, 2547–2557.
14 M. W. Fraaije, J. Wu, D. P. H. M. Heuts, E. W. Hellemond, J. H. L.
Spelberg and D. B. Janssen, Appl. Microbiol. Biotechnol., 2005, 66,
393–400.
3 T. Werpy, et al., Top Value Added Chemicals from Biomass, U. S. D. o.
Energy, Oak Ridge, TN, 2004.
4 K. N. G. Valdehuesa, H. Liu, G. M. Nisola, W.-J. Chung, S. H. Lee and
S. J. Park, Appl. Microbiol. Biotechnol., 2013, 97, 3309–3321;
V. Kumar, S. Ashok and S. Park, Biotechnol. Adv., 2013, 31, 945–961. 15 M. D. Mihovilovic, P. Kapitan and P. Kapitanova, ChemSusChem,
5 J. G. Cannon and J. E. Garst, J. Pharm. Sci., 1975, 64, 1059–1061.
6 M. C. Grubelich, Proc. Int. Pyrotech. Semin., 2002, 29, 133–138.
7 J. H. Helberger, S. Ulubay and H. Civelekoglu, Justus Liebigs Ann.
Chem., 1949, 561, 215–220.
2008, 1, 143–148.
16 M. J. Fink, D. V. Rial, P. Kapitanova, A. Lengar, J. Rehdorf, Q. Cheng,
F. Rudroff and M. D. Mihovilovic, Adv. Synth. Catal., 2012, 354,
3491–3500.
8 R. Noyori, T. Sato and H. Kobayashi, Bull. Chem. Soc. Jpn., 1983, 56, 17 C. M. Clouthier, M. M. Kayser and M. T. Reetz, J. Org. Chem., 2006, 71,
2661–2679.
8431–8437; C. M. Clouthier and M. M. Kayser, Tetrahedron: Asymmetry,
2006, 17, 2649–2653; C. M. Clouthier and M. M. Kayser, J. Mol.
Catal. B: Enzym., 2007, 46, 32–36.
9 M. Snowden, A. Bermudez, D. R. Kelly and J. L. Radkiewicz-Poutsma,
J. Org. Chem., 2004, 69, 7148–7156.
10 G. de Gonzalo, M. D. Mihovilovic and M. W. Fraaije, ChemBioChem, 18 H. M. Dudek, M. J. Fink, A. V. Shivange, A. Dennig, M. D. Mihovilovic,
2010, 11, 2208–2231; H. Leisch, K. Morley and P. C. K. Lau,
Chem. Rev., 2011, 111, 4165–4222; K. Balke, M. Kadow, H. Mallin,
U. Schwaneberg and M. W. Fraaije, Appl. Microbiol. Biotechnol., 2014,
98, 4009–4020.
S. Sass and U. T. Bornscheuer, Org. Biomol. Chem., 2012, 10, 19 F. Rudroff, V. Alphand, R. Furstoss and M. D. Mihovilovic, Org.
6249–6265. Process Res. Dev., 2006, 10, 599–604.
11 H. Iwaki, Y. Hasegawa, S. Wang, M. M. Kayser and P. C. K. Lau, 20 E. Brenna, F. G. Gatti, D. Monti, F. Parmeggiani and A. Sacchetti,
Appl. Environ. Microbiol., 2002, 68, 5671–5684.
12 K. Kostichka, S. M. Thomas, K. J. Gibson, V. Nagarajan and Q. Cheng,
J. Bacteriol., 2001, 183, 6478–6486.
Chem. Commun., 2012, 48, 79–81; M. Bechtold, E. Brenna, C. Femmer,
F. G. Gatti, S. Panke, F. Parmeggiani and A. Sacchetti, Org. Process Res.
Dev., 2012, 16, 269–276.
Chem. Commun.
This journal is ©The Royal Society of Chemistry 2015