Identification and characterization of the furfural and 5-(hydroxymethyl)furfural degradation pathways of Cupriavidus basilensis HMF14

Article abstract of DOI:10.1073/pnas.0913039107

The toxic fermentation inhibitors in lignocellulosic hydrolysates pose significant problems for the production of second-generation biofuels and biochemicals. Among these inhibitors, 5-(hydroxymethyl) furfural (HMF) and furfural are specifically notorious. In this study, we describe the complete molecular identification and characterization of the pathway by which Cupriavidus basilensis HMF14 metabolizes HMF and furfural. The identification of this pathway enabled the construction of an HMF and furfural-metabolizing Pseudomonas putida. The genetic information obtained furthermore enabled us to predict the HMF and furfural degrading capabilities of sequenced bacterial species that had not previously been connected to furanic aldehyde metabolism. These results pave the way for in situ detoxification of lignocellulosic hydrolysates, which is a major step toward improved efficiency of utilization of lignocellulosic feedstock.

Full text of DOI:10.1073/pnas.0913039107

Identification and characterization of the furfural
and 5-(hydroxymethyl)furfural degradation
pathways of Cupriavidus basilensis HMF14
Frank Koopmana,b,c,2,1, Nick Wierckxa,c,d,1,3, Johannes H. de Windea,b,c, and Harald J. Ruijssenaarsa,c,d,3
aBio-Based Sustainable Industrial Chemistry (B-Basic), bDelft University of Technology, Department of Biotechnology, Julianalaan 67, 2628 BC, Delft, The
c
Netherlands; Kluyver Centre for Genomics of Industrial Fermentation, P.O. Box 5057, 2600 GB, Delft, The Netherlands; and dNetherlands Organization
for Applied Scientific Research, Quality of Life, Department of Bioconversion, Julianalaan 67, 2628 BC, Delft, The Netherlands
Edited* by Lonnie O’Neal Ingram, University of Florida, Gainesville, Gainesville, FL, and approved January 19, 2010 (received for review November 11, 2009)
The toxic fermentation inhibitors in lignocellulosic hydrolysates
pose significant problems for the production of second-generation
biofuels and biochemicals. Among these inhibitors, 5-(hydroxy-
methyl)furfural (HMF) and furfural are specifically notorious. In this
study, we describe the complete molecular identification and
characterization of the pathway by which Cupriavidus basilensis
HMF14 metabolizes HMF and furfural. The identification of this
pathway enabled the construction of an HMF and furfural-metabo-
lizing Pseudomonas putida. The genetic information obtained
furthermore enabled us to predict the HMF and furfural degrading
capabilities of sequenced bacterial species that had not previously
been connected to furanic aldehyde metabolism. These results
pave the way for in situ detoxification of lignocellulosic
hydrolysates, which is a major step toward improved efficiency
of utilization of lignocellulosic feedstock.
basilensis HMF14 from soil, by means of enrichment cultures with
HMFas the sole carbon source (12). In the present study, we have
characterized the HMF and furfural degradation pathways of this
bacterium both at the biochemical and the genetic level. The
structural genes were expressed in a heterologous host, Pseudo-
monas putida S12, yielding a strain capable of utilizing HMF and
furfural as sole carbon sources. Using the newly characterized
gene sequences, the furfural or HMF degrading capabilities of
other bacteria could be predicted. The previously undescribed
insights into the furfural and HMF catabolism of C. basilensis
HMF14 and other bacteria may be applied to modify fermenta-
tion hosts to remove furanic aldehydes in situ. This approach by-
passes the requirement for a detoxification pretreatment and
improves the amount of total utilizable carbon in lignocellulosic
hydrolysate. Thus, unique opportunities are created for the ap-
plication of this renewable feedstock for the biotechnological
production of chemicals and fuels.
hydroxymethyl furfural ∣ degradation ∣ inhibitors ∣ lignocellulosic ∣
hydolysate
Results
Identification of Genes Involved in Furfural and HMF Degradation
by Transposon Mutant Screening. A transposon mutant library of
C. basilensis HMF14 was screened for clones that were unable to
grow on furfural and/or HMF. Twenty-five transposon mutants
were selected from 14.000 clones, and the chromosomal DNA
flanking the transposon insertion sites was sequenced to identify
the interrupted genes. Several individual mutants were found to
have a transposon inserted in the same gene, underpinning that
these genes were essential for furfural and HMF metabolism.
Additional primer walking sequencing of up- and downstream
regions of these genes revealed two distinct gene clusters, both
preceded by a LysR-type transcriptional regulator in the reverse
orientation. The nucleotide sequences of these clusters were as-
signed GenBank accession numbers GU556182 and GU556183.
The first cluster contained five genes, designated hmfABCDE,
whereas the other cluster contained four genes: hmfFGH’H
(Fig. 1A). Insertion of a transposon in either of the two clusters
corresponded to two distinct phenotypes. If the hmfABCDE
cluster was interrupted, no growth occurred on either HMF or
furfural, suggesting a—at least partly—shared metabolic pathway
for utilization of furfural and HMF. An insertion in the
ignocellulosic biomass from grasses, wood, agricultural crop
L
residues, and municipal waste provides an abundant and re-
newable source of sugars for the fermentative production of fuels
and base chemicals. Fermentable sugars may be released from
lignocellulosic biomass by a diversity of pretreatment and hydro-
lysis procedures. A major drawback, however, is the formation of
toxic by-products such as organic acids, phenolic compounds, and
furan derivatives.
Furfural (2-furaldehyde, CAS number 98-01-1) and 5-
(hydroxymethyl)furfural (HMF, CAS number 67-47-0) are key
toxic furan derivatives in acid–pretreated lignocellulose hydroly-
sates (1, 2). These furanic aldehydes cause detrimental effects
that are mostly ill-understood at the molecular level, but result
in a decrease of specific growth rates, ethanol yields, and produc-
tivities in both yeasts and bacteria (for review, see refs. 1, 3,
and 4). Efficient detoxification methods are therefore required
to improve the utility of lignocellulosic hydrolysate as a fermen-
tation feedstock (1, 5, 6).
Several methods for the removal of furanic compounds have
been reported, such as ether extraction, alkaline precipitation, or
enzymatically using laccases (for reviews, see refs. 3 and 7).
Bioabatement with microorganisms that degrade furan deriva-
tives may present an important alternative approach (8–12). A
number of microorganisms that metabolize furfural or HMF have
been described (9, 10, 13, 14). A furfural degradation pathway has
been proposed based on enzyme activities in Pseudomonas putida
strains Fu1 (14) and F2 (15). No reports are available on enzymes
involved in HMF degradation, and no pathway for the degrada-
tion of HMF has been proposed to date. Furthermore, no genetic
information on degradation of furanic compounds is available,
except for a number of regulatory and accessory genes that were
recently reported (16).
Author contributions: J.H.d.W. and H.J.R. designed research; F.K. and N.W. performed
research; F.K. and N.W. analyzed data; and F.K., N.W., J.H.d.W., and H.J.R., wrote the paper.
The authors declare no conflict of interest.
*This Direct Submission article had a prearranged editor.
Freely available online through the PNAS open access option.
1F.K. and N.W. contributed equally to this work.
2To whom correspondence should be addressed at: Delft University of Technology,
Julianalaan 67, 2628 15 BC Delft, The Netherlands. E-mail: f.w.koopman@tudelft.nl.
3Present address: BIRD Engineering BV, Westfrankelandsedijk 1, 3115 HG Schiedam, The
Netherlands.
Recently, we isolated the previously undescribed HMF and
furfural-metabolizing Gram-negative bacterium Cupriavidus
This article contains supporting information online at www.pnas.org/cgi/content/full/
0913039107/DCSupplemental.
www.pnas.org/cgi/doi/10.1073/pnas.0913039107
PNAS ∣ March 16, 2010 ∣ vol. 107 ∣ no. 11 ∣ 4919–4924

Fig. 1. Schematic representation of the genetic organization of the hmf genes for furfural and HMF metabolism in C. basilensis HMF14 (A) and other species
(B) that were identified as potential furfural and/or HMF utilizers. Colors correspond to enzyme activities in Fig. 2. Bold numbers (x/y) below arrows indicate the
percentage identity (x) to the corresponding C. basilensis HMF14 protein in a y amino-acid stretch. Orthologous genes were identified by BLASTx homology
searches in the nonredundant protein database of the National Center for Biotechnology Information. Hits for the furfural cluster were defined as relevant
when orthologues for hmfA, B, C, D, and E were present in a single genome, with the hmfA orthologue encoding an enzyme that was at least 50% identical to
HmfA. The same criterion was used to define hmfF and hmfG orthologues, whereas 40% identity to HmfH was used as the criterion for hmfH orthologues.
Numbers in italics indicate genome locus tags of the indicated strain. White arrows depict genes with no assigned metabolic function. (C) Overview of growth
phenotype of tested strains on mineral salts medium with either furfural or HMF (3 mM) as the sole carbon source. ND: not determined.
hmfFGH’H cluster resulted in loss of growth on HMF only. Mu-
tant phenotypes of transposon mutants and BLASTx analysis (17)
of the genes included in the two clusters are summarized
in Table 1.
In P. putida F2 and Fu1, 2-furoyl-CoA is converted into 5-
hydroxy-2-furoyl-CoA by a molybdenum-dependent 2-furoyl-
CoA dehydrogenase. The proteins encoded by hmfABC in
C. basilensis HMF14 correspond to the three subunits that con-
stitute bacterial Mo-dependent dehydrogenases. Functionality of
hmfABC was confirmed by demonstrating furoic-acid dependent
Nitro Blue Tetrazolium reducing activity in cell extracts of
C. basilensis HMF14 (21 ꢀ 5.7 U · g−1) and P. putida S12 coex-
pressing HmfABC and HmfD (42 ꢀ 4 U · g−1). The latter activity
was required to generate 2-furoyl-CoA from 2-furoic acid as the
substrate for HmfABC (Fig. 2A).
In the final steps of the proposed furoic-acid metabolic path-
way of P. putida strains Fu1 and F2, 5-hydroxy-2-furoyl-CoA is
converted into 2-oxoglutarate and CoA via a combination of
spontaneous keto-enol tautomerizations, delactonization and
thioester hydrolysis (Fig. 2B). No enzyme activities had been pre-
viously specified for the lactone and thioester hydrolysis, and no
clear function could be assigned to the remaining gene of the
hmfABCDE cluster, hmfE. Closest HmfE-homologues were an-
notated as proteins of the enoyl-CoA hydratase/isomerase family.
This family encompasses a wide variety of enzymes with diverse
activities and also includes enoyl-CoA hydrolases (18, 19).
This may suggest a role of HmfE in CoA-thioester hydrolysis,
which was tested by incubating cell suspensions of P. putida
Elucidation of the Furfural Catabolic Pathway of C. basilensis HMF14.
The putative enzyme functions encoded by the hmfABCDE
cluster of C. basilensis HMF14 were in good agreement with
the enzyme activities that were reported to constitute the furoic-
acid degradation pathway of Pseudomonas putida strains F2 and
Fu1 (14, 15) (Fig. 2B). The first step of this proposed pathway
involves an acyl-CoA synthetase to produce 2-furoyl-CoA from
2-furoic acid, which activity matches the putative function of
HmfD. The putative function of HmfD was supported by the ac-
cumulation of 2-furoic acid in hmfD-disrupted transposon mu-
tants of C. basilensis HMF14 when cultured in the presence of
furfuryl alcohol or furfural. Furthermore, it was established that
2-furoic acid is the substrate for ATP-dependent CoA ligation by
HmfD. This activity was present in cell extracts of wild-type
C. basilensis HMF14 (316 ꢀ 26.1 U · g−1) and P. putida S12
expressing HmfD (345 ꢀ 24.5 U · g−1), whereas it was absent
in C. basilensis HMF14 transposon mutants in which hmfD
was disrupted.
4920
∣
www.pnas.org/cgi/doi/10.1073/pnas.0913039107
Koopman et al.

Table 1. Growth phenotype of selected C. basilensis HMF14 transposon mutants, and BLASTx analysis and assigned function of genes
involved in furfural and HMF degradation
Growth phenotype
of transposon mutant
Gene
Best BLASTx hit (Acc. No)
Assigned function
MM + citrate MM + furfural MM + HMF
hmfA
+
−
−
Aerobic-type carbon monoxide dehydrogenase
homologue, subunits L and G (YP_726196)
Carbon-monoxide dehydrogenase (YP_293089)
Furoyl-CoA dehydrogenase
large subunit
Furoyl-CoA dehydrogenase
FAD binding subunit
Furoyl-CoA dehydrogenase
2Fe-2S iron sulfer subunit
Furoyl-CoA syntethase
2-oxoglutaroyl-CoA hydrolase
hmfB *
hmfC *
Aerobic-type carbon monoxide dehydrogenase
2Fe-2S iron-sulfur subunit (YP_726194)
Acyl-CoA synthetase (YP_726193)
Enoyl-CoA hydratase/isomerase (YP_293086)
UbiD family decarboxylase (YP_001895811) 2,5-furan-dicarboxylic acid decarboxylase 1
hmfD
hmfE *
hmfF
+
−
−
+
+
+
+
−
−
hmfG
3-octaprenyl-4-hydroxybenzoate
carboxy-lyase (ZP_02881560)
hypothetical protein(YP_293096)
Glucose-methanol-choline
oxidoreductase (YP_001895804)
LysR family transcriptional
regulator (YP_001862747.1)
LysR family transcriptional
regulator (YP_293091.1)
2,5-furan-dicarboxylic acid decarboxylase 2
hmfH’ *
hmfH
NA
+
+
−
HMF/furfural oxidoreductase
hmfR1 *
hmfR2 *
Putative LysR-type transcriptional regulator
Putative LysR-type transcriptional regulator
NA, no assigned function.
*The mutant phenotype was not determined since no transposon mutant was available.
S12 expressing HmfABCDE with 10 mM furoic-acid. Arsenite
(1 mM) was added to inhibit 2-oxoglutarate dehydrogenase. After
overnight incubation 3 mM of 2-oxoglutarate had accumulated,
which is in agreement with previous experiments performed on
P. putida Fu1 (14). No 2-oxoglutarate was formed with control
cells of P. putida S12 expressing HmfABCD. Based on these find-
ings, it was concluded that hmfE encodes a specific 2-oxoglutar-
oyl-CoA thioesterase. This part of the pathway shows strong
similarity to the 3-methylmuconolactone degradation route de-
scribed by Cha et al. (20), in which the CoA-thioester remains
attached after delactonization. The lactone-CoA intermediate
may hydrolyze spontaneously, or lactone hydrolysis is catalyzed
by a nonspecific hydrolase.
In the furfural cluster, no genes were identified for the pathway
upstream of furoic acid, i.e., the oxidations of furfuryl alcohol to
furfural and furoic acid (Fig. 2). Furfuryl alcohol was found to be
oxidized to furfural by an NAD-dependent dehydrogenase activ-
ity. The oxidation of furfural to furoic acid was catalyzed by a
dehydrogenase activity with unknown cofactor that could be re-
placed by phenazine methosulphate/dichlorophenolindophenol
in vitro. Such activities were observed both in C. basilensis
HMF14 and P. putida S12 (Table 2). Therefore, and since none
of the selected C. basilensis HMF14 transposon mutants were
found to be interrupted in genes encoding alcohol or aldehyde
dehydrogenases, it was presumed that redundant, nonspecific de-
hydrogenases were responsible for the upper pathway oxidation
reactions.
Elucidation of the HMF Catabolic Pathway of C. basilensis HMF14. The
HMF degradation pathway of C. basilensis HMF14 was recon-
structed based on putative gene functions of the hmfFGH’H clus-
ter. The hmfFG genes encode two putative decarboxylases of
the UbiD/UbiX type that commonly operate in concert (21, 22).
C. basilensis HMF14 mutants with disrupted hmfFG genes accu-
mulated HMF acid and 2,5-furandicarboxylic acid (FDCA) when
cultured in the presence of HMF, which suggested that these car-
boxylic acids were the substrate for HmfFG. Cell extracts of both
wild-type C. basilensis HMF14 and P. putida S12 expressing
HmfFG formed 2-furoic acid when incubated with FDCA
(6.0 ꢀ 0.1 U · g−1 and 8.6 ꢀ 0.7 U · g−1, respectively). HMF acid
was not decarboxylated to furfuryl alcohol, demonstrating that
Fig. 2. Graphical representation of the HMF (A) and furfural (B) metabolic
pathway in C. basilensis HMF14. B has been adapted from Koenig and
Andreesen (14). Colored hexamers and triangles indicate enzymes with
the following activities: orange hexagon, furfural/HMF oxidoreductase;
red and green triangles, 2,5-furan-dicarboxylic acid decarboxylase; blue tri-
angle, 2-furoyl-CoA synthetase; yellow triangle, furoyl-CoA dehydrogenase;
purple triangle, 2-oxoglutaryl-CoA hydrolase. Colors correspond to the genes
depicted in Fig. 1A. ▪; indicates a lactone hydrolysis that may occur sponta-
neously, or may be catalyzed by a generic lactone hydrolase. Double-pointed
arrows indicate keto-enol tautomerizations. Reactions marked with (*) can
be catalyzed either by HmfH or by (probably nonspecific) dehydrogenases.
ACC; acceptor, which is oxidizedðoxÞ or reducedðredÞ.
Koopman et al.
PNAS
∣
March 16, 2010
∣
vol. 107
∣
no. 11
∣
4921

Table 2. Dehydrogenase activities measured in cell extract of wild-type C. basilensis HMF14 and P. putida S12
Enzyme activity
Substrate
Product
Cofactor
C. basilensis HMF14
P. putida S12
Aldehyde reductase
Furfural
Furfuryl alcohol
Furfural
HMF
HMF alcohol
HMF
Furfuryl alcohol
Furfural
Furoic acid
HMF alcohol
HMF
NADH
NADþ
109 81
28 10.4
245 25
58 7.5
ND
3305 685.4
172 13.1
410 169.5
925 268.9
ND
Alcohol dehydrogenase
Aldehyde dehydrogenase
Aldehyde reductase
Alcohol dehydrogenase
Aldehyde dehydrogenase
PMS/DCPIP
NADH
NADþ
HMF acid
PMS/DCPIP
12 1.5
27 12.8
ND, not determined for lack of commercially available substrate. Activities depicted in U g−1 protein (1 U represents the amount of
protein that converts 1 μmol of substrate per min). PMS/DCPIP: phenazine methosulphate/2,6-dichlorophenol-indophenol. Errors
denote the deviation of the mean.
FDCA was the actual substrate for HmfFG. Thus, HMF degra- S12_fur was obtained, which showed a reproducible growth rate
of 0.3 h−1 on furfural as a sole carbon source with a biomass yield
of 51% (C-mol biomass/C-mol substrate). Wild-type P. putida S12
and the P. putida S12 strains expressing only HmfABCD or
HmfABC both failed to grow on furoic acid, even after prolonged
incubation. These results confirmed that all genes required
for furfural metabolism are located in the furfural cluster
hmfABCDE and that all genes in this cluster are essential for
furfural metabolism, including the hmfE-encoded CoA-thioester
hydrolase.
dation in C. basilensis HMF14 proceeds obligately via its dicar-
boxylic acid form. No decarboxylase activity was observed in a
cell extract of P. putida S12 expressing HmfG only. When HmfF
was expressed as a single enzyme only slight decarboxylase activ-
ity was observed (0.7 ꢀ 0.1 U · g−1), demonstrating that both pro-
teins are required for optimal FDCA decarboxylase activity.
The hmfH gene encodes a putative FAD-dependent oxidore-
ductase. C. basilensis HMF14 mutants with a disrupted hmfH
gene accumulated HMF acid when cultured in the presence of
HMF. Cell extracts of both wild-type C. basilensis HMF14 and
P. putida S12 expressing HmfH formed FDCA when incubated
with HMF acid, confirming that HmfH catalyzes the oxidation
of the HMF-monocarboxylic acid to the dicarboxylic acid form.
No FDCA was formed when oxygen was removed, demonstrating
that HmfH is a true oxidase.
Subsequently, the hmfFGH genes were cloned into P. putida
S12_fur. The resulting strain, P. putida S12_HMF, utilized either
furfural or HMF as the sole carbon source, at a growth rate of
0.23 h−1 and a yield of 40% (C-mol biomass/C-mol substrate).
This growth rate was comparable with the growth rate of
C. basilensis HMF14 on furfural and HMF (0.22 and 0.25 h−1
,
The hmfH’ gene encodes a hypothetical protein with 49%
identity over a stretch of 296 amino acids to a probable extracy-
toplasmic solute receptor of Ralstonia eutropha H16. This gene
may play a role in HMF transport, but a metabolic function
was considered unlikely (23).
respectively). Without the addition of the hmfFGH genes, no
growth was observed on HMF by P. putida S12_fur. Gene hmfH’
was apparently dispensable for growth on HMF, confirming that
the encoded protein has no essential function in HMF metabo-
lism. Nor was the gene essential for HMF transport in P. putida
S12. Thus, also all genes required for the utilization of HMF were
characterized, and their functionality was reconfirmed by func-
tional expression in a heterologous host.
Analogous to the furfural pathway, no specific genes encoding
dehydrogenases were identified for the oxidations in the upper
HMF metabolic pathway leading from HMF alcohol to HMF
and HMF acid. Also these oxidations were concluded to be per-
formed by nonspecific, redundant dehydrogenases whose activ-
ities were observed both in C. basilensis HMF14 and P. putida
S12 (Table 2). However, it was observed that HmfH could also
oxidize HMF, furfural, and furfuryl alcohol to the corresponding
acids. Apparently, this oxidase is essential for the formation of
FDCA from HMF acid but also provides an oxidase alternative
to the nonspecific alcohol and aldehyde dehydrogenases that con-
stitute the upper metabolic pathways for HMF and furfural.
Based on the above observations, the pathway depicted in
Fig. 2A was constructed for HMF catabolism. First, HMF is oxi-
dized to HMF acid, either by nonspecific dehydrogenases or by
HmfH. Subsequently, HMF acid is oxidized to FDCA for which
conversion HmfH is essential. The HMF and the furfural cata-
bolic pathways converge at the level of 2-furoic acid upon decar-
boxylation of FDCA by HmfFG.
Identification of Other Bacteria Capable of Degrading Furfural and
HMF. The sequence of the C. basilensis HMF14 furfural and HMF
catabolic genes were used in BLAST searches for similar se-
quences in publicly available microbial genomes. Remarkably,
orthologous genes were found in a relatively limited group of
Gram-negative bacteria belonging to the genera Cupriavidus,
Burkholderia, Bradyrhizobium, Rhodopseudomonas, Acidiphilium,
Dinoroseobacter, and Methylobacterium (24–30) (Fig. 1). Eight of
these potential furfural or HMF degrading bacteria were selected
and tested for the ability to utilize furfural or HMF as the sole
carbon source (Fig. 1C).
The organization of the genes in the hmfABCDE (“furfural”)
cluster was identical in all species identified by BLAST searches,
indicating that this cluster is highly conserved. Indeed, all tested
bacteria possessing hmfABCDE orthologues utilized furfural as a
sole carbon source (Fig. 1C). The organization of hmfF, hmfG,
and hmfH, on the other hand, was highly diverse. Of these genes,
the hmfH orthologues appeared to be the least conserved, encod-
ing oxidases that are between 68 and 43% identical to HmfH.
Although the presence of hmfFGH orthologues correlated well
with the ability to utilize HMF (Fig. 1C), the percentage identity
of the ill-conserved oxidase to HmfH appeared to be crucial.
Bradyrhizobium japonicum USDA110 utilized HMF, having an
oxidase that was 45% identical to HmfH. By contrast, Burkhol-
deria xenovorans LB400 was unable to utilize HMF although its
oxidase was 44% identical to HmfH. This observation suggests
that the relative high similarity to HmfH may point at HMF-
acid oxidative function. Keeping this in mind, the capability to
Heterologous Expression of the Furfural and HMF Degradation Path-
ways in P. putida S12. Functional characterization of the furfural
and HMF catabolic genes of C. basilensis HMF14 enabled a re-
construction of the complete catabolic pathway for these furanic
compounds. For a final verification of the functionality of the re-
constructed pathway, the encoding genes were expressed in a het-
erologous host, P. putida S12.
First, the furfural cluster hmfABCDE was introduced into
P. putida S12. As expected, the resulting strain, P. putida S12
pJT′hmfABCDE, was able to utilize furoic acid, furfural, and fur-
furyl alcohol as sole carbon sources, although growth was initially
poor. Therefore, strain S12 pJT′hmfABCDE was repeatedly
transferred to a fresh mineral salts medium with furfural as
the sole carbon source. After 10 serial transfers, P. putida strain utilize furfural and/or HMF could be well predicted in other
4922
∣
www.pnas.org/cgi/doi/10.1073/pnas.0913039107
Koopman et al.

bacteria based on the presence of furfural and HMF metabolic
gene clusters.
biosynthetic pathway. The same holds for the oxygen-dependent
HmfH, which currently limits the applicability of the HMF degra-
dation pathway to aerobic hosts. The applicability of the furfural
and HMF pathway could be broadened by employing an alterna-
tive, non-O2-dependent HMF/furfural oxidizing enzyme and a
non-MoCo furoyl-CoA dehydrogenase; this approach is currently
pursued in our laboratory. The identification of the genes respon-
sible for HMF and furfural degradation furthermore enables the
identification of other bacteria capable of degrading furfural and
HMF. Thus, more unique pathways for furfural and HMF meta-
bolism may be expected to emerge from database mining in the
near future. This will extend the possibilities of implementing
these pathways in industrial fermentation hosts to overcome the
furanic aldehyde inhibition connected with the use of lignocellu-
losic hydrolysate as renewable fermentation feedstock.
Discussion
In this paper the HMF and furfural metabolic pathways of
C. basilensis HMF14 were identified, and the genes involved were
isolated and characterized. Previous to the present study only
fragmented knowledge was available on enzymes involved in
furanic aldehyde metabolism (13–16), while the genetic back-
ground on the metabolism of these compounds was not known
to date. Such knowledge is extremely valuable in alleviating
furanic aldehyde inhibition, which is a serious problem in fermen-
tative production of biofuels and (base) chemicals from acid-
pretreated lignocellulosic hydrolysate feedstock.
With respect to furfural metabolism, our findings partially
confirmed a putative route previously proposed for P. putida
strains Fu1 and F2 (14, 15). In addition, a unique enzyme essen-
tial for furfural degradation was identified. This enzyme, encoded
by hmfE, is likely a 2-oxoglutaroyl-CoA-thioester hydrolase,
although it could not be excluded that CoA-thioester hydrolysis
occurs prior to the lactone hydrolysis as suggested previously
(15). No specific lactone hydrolases were found to be essential
for furfural metabolism, suggesting that lactone hydrolysis either
occurs spontaneously or is catalyzed by a nonspecific lactonase.
We have identified and characterized a previously undescribed
pathway for the utilization of HMF. It was demonstrated that the
degradation of HMF proceeds via FDCA. This compound is
decarboxylated to furoic acid, which is further metabolized by
the furfural degradation route. The formation of FDCA from
HMF requires an FAD-dependent oxidoreductase, encoded by
hmfH that shows a mere 68% identity to the closest homologue
in the nonredundant Genbank database. This unique oxidoreduc-
tase was found to be essential for the oxidation of HMF acid to
FDCA, but it can also oxidize HMF alcohol, HMF, furfuryl alco-
hol, and furfural to the corresponding monocarboxylic acid
forms. Thus, HmfH provides an alternative to the nonspecific
dehydrogenases that also perform these upper pathway oxida-
tions in C. basilensis HMF14. Such nonspecific “furanic dehydro-
genases” were also observed in P. putida S12, as well as in other
microorganisms that cannot utilize furanic aldehydes for growth,
probably serving to detoxify furanic aldehydes (31–33).
Materials and Methods
Strains and Culture Conditions. Strains and plasmids used in this study and the
culture conditions are presented in SI Text. Carbon sources were added to the
mineral salts media (mineral salts medium (MM), adapted from ref. 34) as
indicated. Antibiotics were added as required, in the following concentra-
tions: ampicillin (amp), 100 μg∕ml (Escherichia coli); gentamicin, 30 μg∕ml
in Luria broth (LB), 10 μg∕ml in MM (P. putida S12 (35)); kanamycin,
50 μg∕ml and tetracycline, 10 mg∕L (E. coli), and 15 mg∕L (C. basilensis
HMF14 (12)). E. coli was cultured at 37 °C; all other bacteria were routinely
cultured at 30 °C. Shake flask cultures in mineral salts media were performed
in amber Boston bottles in a horizontally shaking incubator. Cultures in LB
were performed in Erlenmeyer flasks.
Transposon Mutagenesis and Mutant Library Screening. A transposon mutant
library of C. basilensis HMF14 was constructed using a modified version
of plasposon pTnModKmO (36), in which the kanamycin marker gene
(36, 37) was replaced with a tetA tetracycline marker flanked by loxP
sites (SI Text) (38, 39). The resulting plasposon pTnModTcO(lox) was intro-
duced in C. basilensis HMF14 by triparental mating using Escherichia coli
pRK2013 as the mobilizing strain (37, 40). C. basilensis HMF14 transposon
mutants were selected on Pseudomonas isolation agar (Difco) supplemented
with 60 mg∕L Tc and transferred to 96-well plates. The mutant library was
screened on solid MM supplemented with 15 mg∕l Tc and 10 mM of either
furfural, HMF or citrate (positive control substrate). Citrateþ, furfural− and/or
HMF− colonies were selected for further study after pure culturing and re-
confirmation of the phenotype.
DNA Techniques. All primers used are displayed in SI Text, and the construction
of expression plasmids is described in SI Text. Maps of the expression plasmids
pJT′mcs and pBT′mcs are presented in SI Text. Genomic DNA, plasmid DNA,
and agarose-trapped DNA fragments were isolated with commercial kits
(QIAGEN). PCR reactions were performed with Accuprime Pfx polymerase
(Invitrogen) according to the manufacturer’s instructions. Plasmid DNA was
introduced into electrocompetent cells using a Gene Pulser electroporation
device (BioRad). Chromosomal DNA flanking transposon insertion sites were
isolated by standard methods (41). Oligonucleotide synthesis and nucleotide
sequencing were performed by Eurofins MWG operon.
In addition to the genes encoding the core enzymes of the fura-
nic aldehyde metabolic pathways, several other genes were iden-
tified in the transposon mutant library screening, or were present
adjacent to the HMF and furfural gene clusters (SI Text). These
genes apparently had an indirect relation with furanic aldehyde
metabolism, encoding proteins with diverse functions that are
putatively involved in transcriptional regulation, transport, stress
tolerance, and cofactor metabolism. Homologues of some of
these genes were recently isolated from P. putida Fu1 as genes
relating to furfural degradation or toxicity (16). The identified
P. putida Fu1 genes also included a lysR-type regulator gene that
was essential for furoic-acid metabolism. This protein encoding
this gene shows 43% identity over a 298 amino acid stretch to
hmfR1 and 47% identity over a 171 amino acid stretch to hmfR2,
which supports their putative function as transcriptional regula-
tors of the HMF and furfural clusters of C. basilensis HMF14.
The characterization of the biochemical degradation routes of
HMF and furfural paves the way for constructing industrial pro-
duction hosts that remove these toxicants from lignocellulosic
hydrolysate in situ. Thus, detoxification pretreatments prior to
fermentation may be omitted, simplifying process setup and
contributing to cost-effectiveness of lignocellulose-based fermen-
tation processes. As demonstrated in the present study, the HMF
and furfural metabolic pathway can be functionally expressed in
P. putida, but the molybdenum-dependent furoyl-CoA dehydro-
genase encoded by hmfABC puts a constraint on the host with
regard to the requirement for a molybdenum cofactor (MoCo)
Analytical Methods. Cell dry weight (CDW) content of bacterial cultures was
determined by measuring optical density at 600 nm (OD600) using a conver-
sion factor per OD600 unit of 0.56 g CDW/L [Biowave Cell Density Meter (WPA
Ltd)] or 1.4 g CDW/L [μQuant MQX200 microplate reader (Biotek) using flat-
bottom 96-well microplates (Greiner)].
HPLC analyses were performed on an Agilent 1100 system equipped with
a diode array detector. Furanic compounds were measured as previously de-
scribed (12). Using an Aminex HDP-87H column (BioRad, 300 × 7.8 mm), 2-Ox-
oglutarate was detected at 210 nm. As eluent, 4 mM H2SO4 was used at a
flow of 0.6 ml∕ min. Alternatively, 2-oxoglutarate was measured in an enzy-
matic assay using glutamate dehydrogenase as described by Trudgill et al.
(15). Protein concentrations were measured using Bradford reagent (Sig-
ma-Aldrich).
Chemicals. FDCA was purchased from Immunosource B.V. 5-Hydroxymethyl-
furoic acid (HMF acid) was purchased from Matrix Scientific. This compound
was found to be highly esterified. Immediately prior to use 14.6 mg of HMF
acid was dissolved in 10 mL of demineralized water and boiled for 2 h in 2 M
H2SO4, cooled, and adjusted to pH 7.0 with NaOH after addition of 50 mM of
Koopman et al.
PNAS
∣
March 16, 2010
∣
vol. 107
∣
no. 11
∣
4923

phosphate buffer. This solution was assumed to be 7.5 mM stock, after
confirmation of deesterification by HPLC analysis. All other chemicals were
purchased from Sigma-Aldrich Chemie B.V.
controls were incubated in headspace vials with a rubber stopper under
nitrogen gas. Prior to starting the reaction, oxygen was stripped from the
reaction mixture with a continuous stream of nitrogen gas.
FDCA decarboxylase activity was determined by incubating 1.38 mL cell
extract, 0.4 mL MM, and 20 μl of 1 mM pyridoxal 5′-phosphate. The reaction
was started by addition of 0.2 mL of 10 mM FDCA. Samples were drawn at set
intervals and analyzed by HPLC. Immediately after sampling, the reaction was
stopped by addition of HCl to a final concentration of 1 M. One unit is de-
fined as the activity catalyzing the conversion of 1 μmol of furoic acid per min
at 30 °C.
Enzyme Assays. Enzyme activities of the furanic aldehyde pathways were
measured in cell extracts of wild-type C. basilensis HMF14 or P. putida S12
transformants expressing the proper enzyme(s). As
a negative control,
wild-type P. putida S12 was used, or a C. basilensis HMF14 transposon mutant
with the appropriate gene disrupted. Cell extracts were prepared by sonica-
tion from 15-fold concentrated late-log phase cultures on MM supplemented
with 12 mM succinic acid. After removing cell debris, the supernatant was
desalted using a PD10 gel filtration column (GE Healthcare). Enzyme activities
of the furfural metabolic pathway were assessed by methods adapted from
previous reports (13,14) (SI Text).
HMF/furfural oxidoreductase activity was determined by incubating cell
extract with furfural, furfuryl alcohol, HMF, or HMF-acid at 30 °C under oxy-
genated conditions. The reaction mixture contained 1.38 mL cell extract,
0.4 mL oxygen-saturated MM, and 20 μl of 2 mM flavin-adenine dinucleotide
(FAD). The reaction was started by addition of 0.2 mL of a 10 mM substrate
stock solution (furfural, furfuryl alcohol, or HMF). Samples were drawn at set
intervals and analyzed by HPLC. Immediately after sampling, the reaction was
stopped by addition of HCl to a final concentration of 1 M. Oxygen-depleted
ACKNOWLEDGMENTS. The authors thank Professor Michael Sadowsky from
the University of Minnesota for providing a culture of B. japonicum USDA110
and Professor Caroline S. Harwood from the University of Washington for
providing a culture of Rhodopseudomonas palustris BisB18. We also thank
Karin Nijkamp for the construction of the pBT′mcs expression vector. This
project was financially supported by the Netherlands Ministry of Economic
Affairs and the B-Basic partner organizations (www.b-basic.nl) through
B-Basic, a public-private NWO-ACTS (Advanced Chemical Technologies for
Sustainability) program. This project was cofinanced by the Kluyver Centre
for Genomics of Industrial Fermentation, which is part of the Netherlands
Genomics Initiative/Netherlands Organisation for Scientific Research.
1. Almeida JRM, Bertilsson M, Gorwa-Grauslund MF, Gorsich S, Liden G (2009) Metabolic
effects of furaldehydes and impacts on biotechnological processes. Appl Microbiol
Biotechnol 82(4):625–638.
2. Heer D, Sauer U (2008) Identification of furfural as a key toxin in lignocellulosic hydro-
lysates and evolution of a tolerant yeast strain. Microb Biotechnol 1(6):497–506.
3. Palmqvist E, Hahn-Hagerdal B (2000) Fermentation of lignocellulosic hydrolysates. II.
Inhibitors and mechanisms of inhibition. Bioresource Technol 74:25–33.
4. Thomsen MH, Thygesen A, Thomsen AB (2009) Identification and characterization of
fermentation inhibitors formed during hydrothermal treatment and following SSF of
wheat straw. Appl Microbiol Biotechnol 83(3):447–455.
5. Martinez A, Rodriguez ME, York SW, Preston JF, Ingram LO (2000) Effects of Ca(OH)(2)
treatments (“overliming”) on the composition and toxicity of bagasse hemicellulose
hydrolysates. Biotechnol Bioeng 69(5):526–536.
6. Palmqvist E, Hahn-Hagerdal B (2000) Fermentation of lignocellulosic hydrolysates. I.
Inhibition and detoxification. Bioresource Technol 74:17–24.
7. Mussatto SI, Roberto IC (2004) Alternatives for detoxification of diluted-acid lignocel-
lulosic hydrolyzates for use in fermentative processes: A review. Bioresource Technol
93(1):1–10.
8. Yu J, Stahl H (2008) Microbial utilization and biopolyester synthesis of bagasse hydro-
lysates. Bioresource Technol 99(17):8042–8048.
9. Nichols NN, Dien BS, Guisado GM, Lopez MJ (2005) Bioabatement to remove inhibitors
from biomass-derived sugar hydrolysates. Appl Biochem Biotechnol 121(1–3):379–390.
10. Lopez MJ, Nichols NN, Dien BS, Moreno J, Bothast RJ (2004) Isolation of microorgan-
isms for biological detoxification of lignocellulosic hydrolysates. Appl Microbiol Bio-
technol 64(1):125–131.
11. Palmqvist E, HahnHagerdal B, Szengyel Z, Zacchi G, Reczey K (1997) Simultaneous
detoxification and enzyme production of hemicellulose hydrolysates obtained after
steam pretreatment. Enzyme Microb Technol 20(4):286–293.
21. Gulmezian M, Hyman KR, Marbois BN, Clarke CF, Javor GT (2007) The role of UbiX in
Escherichia coli coenzyme Q biosynthesis. Arch Biochem Biophys 467(2):144–153.
22. Howlett BJ, Bar-Tana J (1980) Polyprenyl p-hydroxybenzoate carboxylase in flagella-
tion of Salmonella typhimurium. J Bacteriol 143(2):644–651.
23. Antoine R, et al. (2003) Overrepresentation of a gene family encoding extracytoplas-
mic solute receptors in Bordetella. J Bacteriol 185(4):1470–1474.
24. Pohlmann A, et al. (2006) Genome sequence of the bioplastic-producing “Knallgas”
bacterium Ralstonia eutropha H16. Nat Biotechnol 24(10):1257–1262.
25. Chain PSG, et al. (2006) Burkholderia xenovorans LB400 harbors a multi-replicon, 9.73-
Mbp genome shaped for versatility. Proc Natl Acad Sci USA 103(42):15280–15287.
26. Kaneko T, et al. (2002) Complete genomic sequence of nitrogen-fixing symbiotic
bacterium Bradyrhizobium japonicum USDA110. DNA Res 9(6):189–197.
27. Oda Y, et al. (2008) Multiple genome sequences reveal adaptations of
a
phototrophic bacterium to sediment microenvironments. Proc Natl Acad Sci USA
105(47):18543–18548.
28. Harrison AP (1981) Acidiphilium cryptum gen. nov., sp. nov., heterotrophic bacterium
from acidic mineral environments. Int J Syst Bacteriol 31(3):327–332.
29. Biebl H, et al. (2005) Dinoroseobacter shibae gen. nov., sp nov., a new aerobic photo-
trophic bacterium isolated from dinoflagellates. Int
55:1089–1096.
J Syst Evol Microbiol
30. Green PN, Bousfield IJ (1983) Emendation of Methylobacterium Patt, Cole, and
Hanson 1976, Methylobacterium rhodinum (Heumann 1962) comb. nov. corrig.;
Methylobacterium radiotolerans (Ito and Iizuka 1971) comb. nov. corrig.; and Methy-
lobacterium mesophilicum (Austin and Goodfellow 1979) comb. nov. Int J Syst Bacter-
iol 33(4):875–877.
31. Boopathy R, Bokang H, Daniels L (1993) Biotransformation of furfural and 5-hydroxy-
methyl furfural by enteric bacteria. J Ind Microbiol Biotechnol 11:147–150.
32. Okuda N, Soneura M, Ninomiya K, Katakura Y, Shioya S (2008) Biological detoxifica-
tion of waste house wood hydrolysate using Ureibacillus thermosphaericus for
bioethanol production. J Biosci Bioeng 106(2):128–133.
33. Petersson A, et al. (2006) A 5-hydroxymethyl furfural reducing enzyme encoded by the
Saccharomyces cerevisiae ADH6 gene conveys HMF tolerance. Yeast 23(6):455–464.
34. Hartmans S, Smits JP, van der Werf MJ, Volkering F, de Bont JA (1989) Metabolism of
styrene oxide and 2-phenylethanol in the styrene-degrading Xanthobacter Strain
124X. Appl Environ Microbiol 55(11):2850–2855.
35. Hartmans S, van der Werf MJ, de Bont JA (1990) Bacterial degradation of styrene
involving a novel flavin adenine dinucleotide-dependent styrene monooxygenase.
Appl Environ Microbiol 56(5):1347–1351.
12. Wierckx N, Koopman F, Bandounas L, de Winde JH, Ruijssenaars HJ (2009) Isolation and
characterization of Cupriavidus basilensis HMF14 for biological removal of inhibitors
from lignocellulosic hydrolysate. Microb Biotechnol 10.1111/j.1751-7915.2009.
00158.x.
13. Koenig K, Andreesen JR (1989) Molybdenum involvement in aerobic degradation of
2-furoic acid by Pseudomonas putida Fu1. Appl Environ Microbiol 55(7):1829–1834.
14. Koenig K, Andreesen JR (1990) Xanthine dehydrogenase and 2-furoyl-coenzyme A
dehydrogenase from Pseudomonas putida Fu1: Two molybdenum-containing dehy-
drogenases of novel structural composition. J Bacteriol 172(10):5999–6009.
15. Trudgill PW (1969) Metabolism of 2-furoic acid by Pseudomonas F2. Biochem
113(4):577–587.
J
36. Dennis JJ, Zylstra GJ (1998) Plasposons: Modular self-cloning minitransposon deriva-
tives for rapid genetic analysis of gram-negative bacterial genomes. Appl Environ
Microbiol 64(7):2710–2715.
37. Wierckx NJP, Ballerstedt H, de Bont JAM, Wery J (2005) Engineering of solvent-tolerant
Pseudomonas putida S12 for bioproduction of phenol from glucose. Appl Environ
Microbiol 71(12):8221–8227.
16. Nichols NN, Mertens JA (2008) Identification and transcriptional profiling of Pseudo-
monas putida genes involved in furoic acid metabolism. FEMS Microbiol Lett
284(1):52–57.
17. Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ (1990) Basic local alignment search
tool. J Mol Biol 215(3):403–410.
18. Hawes JW, et al. (1996) Primary structure and tissue-specific expression of human
beta-hydroxyisobutyryl-coenzyme A hydrolase. J Biol Chem 271(42):26430–26434.
19. Holden HM, Benning MM, Haller T, Gerlt JA (2001) The crotonase superfamily: Diver-
gently related enzymes that catalyze different reactions involving acyl coenzyme A
thioesters. Acc Chem Res 34(2):145–157.
20. Cha CJ, Cain RB, Bruce NC (1998) The modified beta-ketoadipate pathway in
Rhodococcus rhodochrous N75: Enzymology of 3-methylmuconolactone metabolism.
J Bacteriol 180(24):6668–6673.
38. Sauer B, Henderson N (1988) Site-specific DNA recombination in mammalian-cells by
the Cre recombinase of bacteriophage-P1. Proc Natl Acad Sci USA 85(14):5166–5170.
39. Sternberg N, Hamilton
D (1981) Bacteriophage P1 site-specific recombination.
1. Recombination between loxP sites. J Mol Biol 150(4):467–486.
40. Ditta G, Stanfield S, Corbin D, Helinski DR (1980) Broad host range DNA cloning system
for gram-negative bacteria: Construction of a gene bank of Rhizobium meliloti. Proc
Natl Acad Sci USA 77:7347–7351.
41. Ausubel FM, et al. (1987) Current Protocols in Molecular Biology (Greene Publishing
Association, New York).
4924
∣
www.pnas.org/cgi/doi/10.1073/pnas.0913039107
Koopman et al.