Enzyme cascade reactions: Synthesis of furandicarboxylic acid (FDCA) and carboxylic acids using oxidases in tandem

Article abstract of DOI:10.1039/c5gc00707k

A one-pot tandem enzyme reaction using galactose oxidase M_3-5 and aldehyde oxidase PaoABC was used to convert hydroxymethylfurfural (HMF) to the pure bioplastics precursor FDCA in 74% isolated yield. A range of alcohols was also converted to carboxylic acids in high yield under mild conditions.

Full text of DOI:10.1039/c5gc00707k

Green Chemistry
View Article Online
View Journal
COMMUNICATION
Enzyme cascade reactions: synthesis
of furandicarboxylic acid (FDCA) and
Cite this: DOI: 10.1039/c5gc00707k
carboxylic acids using oxidases in tandem†
Received 1st April 2015,
Accepted 21st April 2015
Shane M. McKenna,a Silke Leimkühler,b Susanne Herter,c Nicholas J. Turner*c and
DOI: 10.1039/c5gc00707k
Andrew J. Carnell*a
www.rsc.org/greenchem
A one-pot tandem enzyme reaction using galactose oxidase M3–5 furanic chemicals.6 Obtaining 2 from biomass at low cost will
and aldehyde oxidase PaoABC was used to convert hydroxy- be essential to allow a paradigm shift in green manufacturing
methylfurfural (HMF) to the pure bioplastics precursor FDCA in although this target has yet to be realized. HMF 1 consists of a
74% isolated yield. A range of alcohols was also converted to furan ring with 2,5-disposed aldehyde and hydroxymethyl
carboxylic acids in high yield under mild conditions.
functional groups. It can be synthesized from glucose/fructose
by dehydration7 in high yield although currently only with con-
tinuous removal of water or extraction into non aqueous sol-
vents or ionic liquids, due to instability in water at high
temperature.7d In order to obtain 2 from 1 a 6 electron oxi-
dation is required. Numerous metal catalysts and nano-
particles have been employed such as Au–TiO2,8 Au–C
modified with Pd,9 Au–hydrotalicite,10 Pt–C,11 Au/TiO2,12 Pt/
ZrO2,13 however these reactions require high pressure/tempera-
ture and additives which decreases the sustainability for
manufacturing considerably. Thus there remain a number of
challenges for the steps from cellulose to FDCA and their
integration.
A catalytic system that uses O2 from air and produces water
as the only by-product would contribute to establishing a
green and sustainable process for conversion of HMF 1 to
FDCA 2 (Scheme 1).14 To this end, aerobic Pt nanoparticles15
and gold systems16 have been developed but usually require
high pH which can partially decompose the unstable HMF.
Biocatalytic reactions offer many benefits in the context of
green chemistry since they can be performed under mild con-
ditions using a biodegradable catalyst.17 The synthesis of 2
Green chemistry encompasses a set of principles that can be
applied in designing sustainable chemical processes. This
includes use of renewable raw materials, elimination of waste
and avoiding the use of toxic and hazardous reagents and sol-
vents.1 With the rapid growth of the world population and con-
tinuing depletion of petroleum reserves, green approaches
using renewable resources for the production of chemicals will
be required. Lignocellulosic biomass is an abundant, inexpen-
sive and sustainable resource from which platform chemicals
can be derived. 5-Hydroxymethylfurfural (HMF) 1 is derived
from cellulose via dehydration of glucose and fructose. Due to
the instability of HMF, its oxidized and more stable form
furan-2,5-dicarboxylic acid (FDCA) 2 is listed as one of twelve
sugar-based platform chemicals of interest by the American
DOE.2
Polyethylene terephthalate (PET) makes up 5.9% of the
global plastics industry with approximately 15 million tons per
year being manufactured. It is considered that bio-based 2
could replace terephthalic acid in this and related co-polymers
which would be a substantial step towards sustainable plastics
manufacture.3 Moreover, 2 is a building block2b and can be
used to synthesize other polyesters,4 polyamides5 and valuable
aDepartment of Chemistry, University of Liverpool, Crown Street, Liverpool, UK.
E-mail: acarnell@liv.ac.uk; Fax: +44 (0)151 7943500; Tel: +44 (0)151 7943534
bInstitute of Biochemistry and Biology, University of Potsdam, D-14476 Potsdam,
Germany. E-mail: sleim@uni-potsdam.de; Fax: +44 331/977-5128;
Tel: +49 331/977-5603
cSchool of Chemistry, Manchester Institute of Biotechnology, University of
Manchester, 131 Princess Street, Manchester, M1 7DN, UK.
E-mail: nicholas.turner@manchester.ac.uk; Fax: +44 (0)161 2751311;
Tel: +44 (0)161 3065173
†Electronic supplementary information (ESI) available: HPLC data and NMR
spectra. See DOI: 10.1039/c5gc00707k
Scheme 1 Desired route to FDCA from biomass derived sugars.
This journal is © The Royal Society of Chemistry 2015
Green Chem.

View Article Online
Communication
Green Chemistry
from HMF has been previously demonstrated using a recombi- commercially available E. coli XDH, with which we had pre-
nant whole cell biotransformation.18 However the need for a viously shown oxidation of benzylic aldehydes, was only active
continuous carbon source feed and low product recovery of the with 1. The Rhodococcus capsulatus xanthine dehydrogenase
polar FDCA from the cell biomass limited the potential of this (XDH) single variant E232V24 and double mutant XDH E232 V/
process. Bioconversions using isolated enzymes can proceed at R310 showed activity against all three substrates; however
significantly higher substrate concentration and combine these variants possess very low reactivity with oxygen from air
higher productivity with a lower water usage. There are only a and require an exogenous electron acceptor such as DCPIP for
few examples of bioconversions of 1 to 2 in the literature. A high conversions. One of us25 recently reported an E. coli peri-
galactose oxidase mutant and peroxidase were used in a plasmic aldehyde oxidase (PaoABC) which uses oxygen as the
tandem reaction to give 46% yield of impure 2 and Caldario- terminal electron acceptor. PaoABC is a 135 kDa heterotri-
myces fumago chloroperoxidase has been reported to incomple- meric enzyme with a large (78.1 kDa) molybdenum cofactor
tely oxidize 1; both processes use low substrate concentrations (Moco)-containing PaoC subunit, a medium (33.9 kDa) FAD-
(1 mM), require addition of peroxide and result in separation containing PaoB subunit, and a small (21.0 kDa) [2Fe–2S]-con-
issues.19 Recently, an FAD-dependent HMF oxidase has been taining PaoA subunit.25,26 A variety of enzymes have evolved to
reported to fully oxidize 1 to 2 using molecular oxygen, metabolize aldehydes to less reactive intermediates27 and it is
although at low substrate concentration (2–4 mM).20 Aryl believed that PaoABC plays a role in the detoxification of aro-
alcohol oxidase (AAO) was used to convert 1 (3 mM) to 5-for- matic aldehydes.25 PaoABC was able to oxidize all three sub-
mylfuran-2-carboxylic acid (FFCA) 5, then in a second much strates 1, 4 and 5 and hence this enzyme was selected as the
slower step an unspecific peroxygenase (UPO) converted 5 candidate biocatalyst for the aldehyde oxidation step.
to 2.21
Recently we have shown that in vitro cascades employing
A test cascade reaction was set up using 1 with both GOase
3–5 and PaoABC present to identify if 2 could be produced. At
M
two isolated oxidases, GOase M3–5 and xanthine dehydrogen- 10 mM HMF concentration, we were pleased to observe almost
ase (XDH) quantitatively yield aromatic carboxylic acids start- complete conversion (97%) to 2 after 1 h, with key intermedi-
ing from benzylic alcohols via the intermediate aldehyde.22 We ates being 4 and 5 (Table S2, entry 1, Fig. S7†). We then investi-
therefore postulated that this methodology could be adopted gated the effect of increasing the concentration of 1. At 20 mM
for the synthesis of 2 from 1. In the present work we demon- 1, after 1 h 2 (44%) was formed along with 3 (50%) (Fig. 1A,
strate a preparative scale (74% isolated yield, 50–100-fold
greater [S] than previously reported) for an enzymatic synthesis
of 2 from 1 while also expanding the scope for alcohol oxi-
dations under mild conditions. For an enzymatic cascade we
envisaged using galactose oxidase M3–5, plus a suitable
xanthine oxidoreductase (XOR) for sequential oxidations,
ideally in one pot (Scheme 2). One of us previously developed
variants of F. graminearum galactose oxidase (GOase), such as
23
GOase M3–5
,
that shows a remarkable ability to oxidize sec-
ondary and primary benzylic alcohols with H2O2 as the only
by-product.
Initial studies revealed that 1 was indeed a good substrate
for GOase M3–5 yielding the dialdehyde 4. The XOR enzyme
should be able to oxidize the aldehyde groups in 4 and 5, both
of which are intermediates (Scheme 2). Four molybdenum-
dependent XOR enzymes were chosen to screen for activity
against the available substrates 1, 4 and 5 (Table S1†). The
Fig. 1 Enzyme cascades for conversion of HMF (1) with (A) dual com-
bined enzymes (GOase M3–5 + PaoABC) ([HMF] = 20 mM) and (B) one-
pot sequential reaction ([HMF] = 50 mM).
Scheme 2 Possible intermediates en route from HMF 1 to FDCA 2
using two enzymes: galactose oxidase M3–5 and XOR.
Green Chem.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Green Chemistry
Communication
Table S2,† entry 2), then after a further 4 h, a small amount of possibility that a small fraction of 5 may be produced either
3 was slowly converted to give a final conversion of 55% of 2. directly by GOase M3–5 or by spontaneous oxidation by
The time course shows that formation of 2 occurs mainly via 4 residual H2O2, as a low level of 5 is detected before the
and 5 and that 3 is a poor substrate for GOase M3–5. It is also addition of PaoABC (Fig. 1B).
clear that PaoABC is responsible for the formation of 3 at this
substrate concentration.
Turning our attention back to the combined dual enzyme
reaction, we considered that a pH drop might limit oxidation
Production of 3 could be avoided by using a sequential, of the hydroxymethyl acid 3 by GOase M3–5. However, oxi-
stepwise process in which the GOase M3–5 conversion of 1 to 4 dation at [S] = 50 mM 1 with GOase M3–5 and PaoABC com-
was allowed to run to completion prior to addition of the bined at high buffer concentration resulted in predominantly
PaoABC enzyme (Fig. 1B, Table S2,† entries 3–10). This step- 3 after 3 h, which was slowly converted to 5. The latter was
wise reaction furnished the required 2 as the only oxidation then rapidly oxidized to 2 and after 17 h, 56% of 2 was present
product, cleanly and with 100% conversion. Oxidases such as (Fig. S9†).
GOase and PaoABC produce H2O2 as a byproduct and it was
Having established a tandem process for conversion of 1 to
found necessary to remove the peroxide by addition of cata- 2, we further explored the substrate specificity of PaoABC for
lase. Catalase converts H2O2 into O2 and thus provides extra oxidation of a wider range of activated and unactivated
equivalents of O2 for the reaction, while also protecting the aldehydes.
enzymes from oxidative damage.
Ten selected aldehyde substrates were shown to be active
On increasing the starting concentration of HMF (1) further with PaoABC. Subsequent combination with GOase M3–5
we found that sufficient buffer capacity to control pH was resulted in the successful conversion of the corresponding
important to enable the reaction to reach completion primary alcohols to carboxylic acids 6–15 (Table 1) with, in
(Table S2†). During the optimisation of the conversion of 4 to most cases, quantitative conversions at 10 mM substrate con-
2 with PaoABC it was noted that with [S] = 100 mM, not sur- centration. A time course study for the conversion of phenyl-
prisingly a drop in pH was occurring (pH < 5, Table S3,† entry ethanol (10 mM) to phenylacetic acid 8 revealed no aldehyde
7), which was below the optimum pH (6–8) of PaoABC, as intermediate, indicating that the second PaoABC step was
measured with m-anisaldehyde as substrate (Fig. S2†). We extremely rapid (Fig. S10†). However, increasing the substrate
believe that this drop in pH is responsible for the reduced con- concentration to [S] = 30 mM caused problems for the GOase
version, rather than hydrate formation from 5 since although
the dialdehyde 4 does form a hydrate at pH 5–8, the aldehyde-
acid 5 does not (Fig. S3 and S4†). At 100 mM 1, an additional
portion of catalase was required with the addition of PaoABC
(Table S2† entries 8 and 9), possibly due to the extensive reac-
tion time for the first oxidation (>16 h), during which time
catalase may be deactivated. With our optimized conditions in
hand, the preparative scale oxidation of 1 was then realized
(Table S2,† entry 10). Isolation of 2 by crystallization has been
successful after whole cell biotransformations,18 however the
removal of biomass and numerous extractions into organic sol-
vents reduce sustainability. Using our in vitro cascade system,
no organic solvents were required. Heat treatment of the solu-
tion to precipitate the protein, centrifugation, acidification
and filtration is all that was required to obtain pure 2 in
74% isolated yield (ESI 8.0). In the 50 mM sequential reac-
tion (Fig. 1B) the initial oxidation of 1 was found to be the
slower step. On addition of PaoABC there is a rapid oxi-
dation of 4 to 5. It is noteworthy that 2 is not produced
rapidly until all of 4 is oxidized to 5. The dialdehyde 4 is a
better substrate for PaoABC, possibly because its hydrate,
which forms rapidly in buffer (Fig. S3†), is the true sub-
strate. In the sequential reaction, the whole oxidation
process is complete after 8 h. The hydrate, rather than the
aldehyde itself, has recently been reported to be the catalyti-
cally active substrate for other oxidase enzymes, including
HMF oxidase,20 AAO and other enzymes.21,28 In each case
the final aldehyde-acid (5) is a poor substrate because the
equilibrium for hydrate formation is highly unfavourable. In
our two-step sequence we cannot completely exclude the
Table 1 Scope of the GOase M3–5–PaoABC enzymatic cascadea
Product
Conversion
>99%
Product
Conversion
>99%
>99%
>99%
>99%
>99%
81%
>99%
>99%
50%
a Reaction conditions: 103 µL GOase M3–5 (3 mg mL−1), 5 µL PaoABC
(13.2 mg mL−1), 33 µL catalase (1.3 mg mL−1), 3 µL of substrate (1 M
in MeCN 10 mM final concentration) in 50 mM pH 7.6 potassium
phosphate buffer (159 µL), 37 °C shaking overnight.
This journal is © The Royal Society of Chemistry 2015
Green Chem.

View Article Online
Communication
Green Chemistry
M3–5 step. For example, using 3-phenylbutan-1-ol required natant was then cooled to 0 °C and concentrated HCl was
large quantities of GOase to obtain high conversions to 14 added dropwise until a precipitate formed. The solution was
(Table S6†). PaoABC showed a remarkable tolerance of high then centrifuged and the supernatant removed and the pellet
substrate concentration as no aldehyde intermediate could be washed with 1 M HCl. The pellet was dissolved in acetone and
identified by NMR. Since GOase M3–5 currently represents the then concentrated in vacuo three times yielding 2 as a pale
limiting activity it will be necessary to engineer GOase variants yellow solid (35 mg, 0.22 mmol, 74% yield).
that are more tolerant to higher alcohol concentrations. The
GOase M3–5–PaoABC one-pot conversion provides a highly
green and direct method for the conversion of alcohols to car-
boxylic acids and compares favourably with the current
Acknowledgements
approach using a chemocatalytic Ru-pincer complex where
The research leading to these results received support from
reactions are run with 1 equivalent of NaOH in refluxing
the Engineering and Physical Sciences Research Council
water.29
(EPSRC, to SMM and AJC), the Innovative Medicines Initiative
Joint Undertaking under the grant agreement no. 115360
(CHEM21, to SH & NJT) and The Royal Society Wolfson Merit
Award (to NJT).
Conclusions
In summary we have developed a promising tandem cascade
reaction using two oxygen-dependent enzymes, galactose
oxidase M3–5 and aldehyde oxidase PaoABC, that results in
Notes and references
high conversion of 1 to 2 at ambient temperature and near
neutral pH. The substrate concentration of 1 (100 mM) is the
highest reported for an enzyme-based process and uses
enzymes that do not require the addition of diffusible cofac-
1 R. A. Sheldon, I. Arends and U. Hanefeld, Green Chemistry
and Catalysis, Wiley-VCH, 2007.
2 (a) A. Boisen, T. B. Christensen, W. Fu, Y. Y. Gorbanev,
tors. Key the success of this tandem cascade compared with
T. S. Hansen, J. S. Jensen, S. K. Klitgaard, S. Pedersen,
HMF oxidase20 is the ability of PaoABC to efficiently catalyse
A. Riisager, T. Ståhlberg and J. M. Woodley, Chem. Eng. Res.
the final oxidation step of 5 to 2. Whereas HMF oxidase
Des., 2009, 87, 1318–1327; (b) Top Value added Chemicals
requires the hydrate of 5, PaoABC appears to operate on the
from Biomass, ed. T. Werpy and G. Peterson, US Department
aldehyde. Further development of the enzymes and reaction
of Energy, 2004.
conditions, particularly with the aim of interfacing with
3 (a) B. Saha, S. Dutta and M. M. Abu-Omar, Catal. Sci.
Technol., 2012, 2, 79; (b) A. J. J. E. Eerhart, A. P. C. Faaij and
M. K. Patel, Energy Environ. Sci., 2012, 5, 6407.
4 (a) E. Gubbels, L. Jasinsta-Walc and C. E. Koning, J. Polym.
Sci., Part A: Polym. Chem., 2013, 51, 890–893;
(b) R. Storbeck and M. Ballauff, Polymer, 1993, 34, 5004–
5006; (c) J. A. Moore and J. E. Kelly, Macromolecules, 1978,
11, 568–573.
upstream reactions to produce HMF from cellulose could
provide an extremely green and economic route to highly pure
FDCA (2). We have also demonstrated a one-pot, single stage
tandem cascade for the quantitative conversion of 10 alcohols
directly to carboxylic acids, including unactivated examples.
The conditions are green and compare favorably with state-of-
the-art chemocatalytic methods.
5 S. Rastogi, B. A. J. Noordover, Y. S. Deshmukh and C. H. R.
M. Wilsens, Macromolecules, 2014, 47, 6196–6206.
6 C. Moreau, M. N. Belgacem and A. Gandini, Top. Catal.,
Representative procedure for one-pot
tandem enzyme reaction for
preparation of FDCA (2)
2004, 27, 11–30.
7 (a) J. N. Chheda, G. W. Huber and J. A. Dumesic, Angew.
Chem., Int. Ed., 2007, 46, 7164–7183; (b) X. Tong, M. Li,
N. Yan, Y. Ma, P. Dyson and X. Li, Catal. Today, 2011, 175,
524–527; (c) A. Corma, S. Iborra and A. Velty, Chem. Rev.,
2007, 107, 2411–2502; (d) P. Lanzafame, G. Centi and
S. Perathoner, Catal. Today, 2014, 234, 2–12.
8 Y. Y. Gorbanev, S. K. Klitgaard, J. M. Woodley,
C. H. Christensen and A. Riisager, ChemSusChem, 2009, 2,
672–675.
HMF (1) (38 mg, 0.3 mmol, final concentration = 100 mM) and
catalase (0.33 mL of a 3.3 mg mL−1 solution) were added to
potassium phosphate buffer (400 mM pH 7) (1.09 mL) and
MeCN (0.03 mL). GOase M3–5 (1.5 mL of a 3.3 mg mL−1 solu-
tion) was then added and the reaction shaken at 37 °C for 10 h
in a shaking incubator. After this time, another portion of
catalase (0.33 mL of a 3.3 mg mL−1 solution) was added along
with PaoABC (0.05 mL of a 13.2 mg mL−1 solution) and the
reaction shaken for a further 5 h in the incubator. The pH was
9 L. Prati, A. Villa, M. Schiavoni, S. Campisi and G. M. Veith,
ChemSusChem, 2013, 6, 609–612.
carefully monitored and adjusted to pH 7 with 1 M NaOH. 10 N. K. Gupta, S. Nishimura, A. Takagaki and K. Ebitani,
After this time the reaction was heated to 80 °C for 5 minutes Green Chem., 2011, 4, 824.
and left to cool. The solution containing denatured protein 11 S. E. Davis, B. N. Zope and R. J. Davis, Green Chem., 2012,
was centrifuged and the supernatant removed. The super-
1, 143.
Green Chem.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Green Chemistry
Communication
12 M. A. Lilga, R. T. Hallen and M. Gray, Top. Catal., 2010, 53,
1264–1269.
R. Ullrich, M. Hofrichter and A. T. Martínez, FEBS J., 2015,
DOI: 10.1111/febs.13177.
13 Y. Y. Gorbanev, S. Kegnæs and A. Riisager, Top. Catal., 22 B. Bechi, S. Herter, S. McKenna, C. Riley, S. Leimkuhler,
2011, 54, 1318–1324.
N. J. Turner and A. J. Carnell, Green Chem., 2014, 16, 4524–
14 (a) T. Mallat and A. Baiker, Chem. Rev., 2004, 104,
4529.
3037–3058; (b) K. Kaneda, K. Ebitani, T. Mizugaki and 23 (a) F. Escalettes and N. J. Turner, ChemBioChem, 2008, 9,
K. Mori, Bull. Chem. Soc. Jpn., 2006, 79, 981–1016;
(c) B.-Z. Zhan and A. Thompson, Tetrahedron, 2004, 60,
2917–2935.
857–860; (b) B. Yuan, A. Page, C. P. Worrall, F. Escalettes,
S. C. Willies, J. J. W. McDouall, N. J. Turner and J. Clayden,
Angew. Chem., Int. Ed., 2010, 49, 7010–7013.
15 S. Siankevich, G. Savoglidis, Z. Fei, G. Laurenczy, 24 S. Schumann, M. Terao, E. Garattini, M. Saggu, F. Lendzian,
D. T. L. Alexander, N. Yan and P. J. Dyson, J. Catal., 2014,
315, 67–74.
P. Hildebrandt and S. Leimkühler, PLoS One, 2009, 4(4),
e5348.
16 (a) A. S. K. Hashmi and G. J. Hutchings, Angew. Chem., Int. 25 M. Neuman, G. Mittelstadt, C. Iobbi-Nivol, M. Saggu,
Ed., 2006, 45, 7896–7936; (b) A. Corma and H. Garcia,
Chem. Soc. Rev., 2008, 37, 2096–2126.
F. Lendzian, P. Hildebrandt and S. Leimkühler, FEBS J.,
2009, 276, 2762–2774.
17 B. M. Nestl, S. C. Hammer, B. A. Nebel and B. Hauer, 26 (a) M. Neuman and S. Leimkühler, Biochem. Res. Int., 2011,
Angew. Chem., Int. Ed., 2014, 53, 3070–3095.
18 F. Koopman, N. Wierckx, J. H. de Winde and
1–11; (b) C. Iobbi-Nivol and S. Leimkühler, Biochim.
Biophys. Acta, 2013, 1086–1101.
H. J. Ruijssenaars, Bioresour. Technol., 2010, 101, 6291– 27 (a) T. Yao and S. Handa, Anal. Sci., 2003, 19, 767;
6296.
(b) U. Wollenberger, S. Leimkühler, M. Neumann-Schaal
and A. Badalyan, Electroanalysis, 2013, 25, 101–108.
19 (a) L. Kalum, M. D. Morant, H. Lund, J. Jensen,
I. Lapainaite, N. H. Soerensen, S. Pedersen, 28 (a) P. Ferreira, A. Hernández Ortega, B. Herguedas,
L. H. Ostergaard and F. Xu, Patent WO 2014015256 A2,
2014; (b) P. D. Hanke, Patent WO2009/023174, 2009;
(c) R. A. Sheldon and F. Van Rantwik, J. Carbohydr. Chem.,
1997, 16, 299–309.
J. Rencoret, A. Gutiérrez, M. J. Martínez, J. Jimenez-
Barbero, M. Medina and A. T. Martinez, Biochem. J., 2010,
425, 585–593; (b) E. W. van Hellemond, L. Vermote,
W. Koolen, T. Sonke, E. Zandvoort, D. P. Heuts,
D. B. Janssen and M. W. Fraaije, Adv. Synth. Catal., 2009,
351, 1523–1530.
20 W. P. Dijkman, D. E. Groothuis and M. W. Fraaije, Angew.
Chem., Int. Ed., 2014, 53, 1–5.
21 J. Carro, P. Ferreira, L. Rodríguez, A. Prieto, A. Serrano, 29 E. Balaraman, E. Khaskin, G. Leitus and D. Milstein, Nat.
B. Balcells, A. Arda, J. Jiménez-Barbero, A. Gutiérrez,
Chem., 2013, 5, 122–125.
This journal is © The Royal Society of Chemistry 2015
Green Chem.