Creating a more robust 5-hydroxymethylfurfural oxidase by combining computational predictions with a novel effective library design
© The Author(s) 2018
Received: 6 December 2017
Accepted: 14 February 2018
Published: 1 March 2018
HMF oxidase (HMFO) from Methylovorus sp. is a recently characterized flavoprotein oxidase. HMFO is a remarkable enzyme as it is able to oxidize 5-hydroxymethylfurfural (HMF) into 2,5-furandicarboxylic acid (FDCA): a catalytic cascade of three oxidation steps. Because HMF can be formed from fructose or other sugars and FDCA is a polymer building block, this enzyme has gained interest as an industrially relevant biocatalyst.
To increase the robustness of HMFO, a requirement for biotechnological applications, we decided to enhance its thermostability using the recently developed FRESCO method: a computational approach to identify thermostabilizing mutations in a protein structure. To make this approach even more effective, we now developed a new and facile gene shuffling approach to rapidly combine stabilizing mutations in a one-pot reaction. This allowed the identification of the optimal combination of seven beneficial mutations. The created thermostable HMFO mutant was further studied as a biocatalyst for the production of FDCA from HMF and was shown to perform significantly better than the original HMFO.
The described new gene shuffling approach quickly discriminates stable and active multi-site variants. This makes it a very useful addition to FRESCO. The resulting thermostable HMFO variant tolerates the presence of cosolvents and also remained thermotolerant after introduction of additional mutations aimed at improving the catalytic activity. Due to its stability and catalytic efficiency, the final HMFO variant appears to be a promising candidate for industrial scale production of FDCA from HMF.
Since HMFO has the potential to become an important biocatalytic tool, development of a robust variant is essential. An enzyme with a higher thermostability will be more suitable for use under harsh reaction conditions such as elevated temperatures and the presence of organic cosolvents . Moreover, a stable enzyme is the ideal template for further enzyme engineering efforts aimed to improve or change its catalytic properties . Several methods have been designed to enhance the (thermo)stability of enzymes . Completely random approaches like directed evolution can be effective but are time-consuming and require high-throughput screening methods .
Rational design of stabilizing mutations is still challenging due to the fact that it is too complicated to accurately explain the effects of a mutation in terms of ΔΔH and ΔΔS . Yet, state-of-the-art computational methods have evolved to such a level that they have predictive value in selecting putative thermostabilizing mutations. This can be used as input for the design of relatively small mutant libraries for screening in vitro . We have recently developed a framework for rapid enzyme stabilization by computational libraries (FRESCO) which is a computationally assisted method that includes predicting a large number of independent stabilizing mutations that are in silico screened to define a relatively small set of mutations that need to be tested experimentally [18, 19]. This approach has demonstrated its validity with different enzymes, leading to high T m app improvements of up to 35 °C [18, 20–22]. The initial steps of the FRESCO strategy consist of computational and visual selections based on free energy predictions and molecular dynamic (MD) simulations of single point mutations. This is followed by an in vitro phase to identify variants with significantly improved stability, and finally, the combination of the selected mutations should result in a highly stable enzyme variant. FRESCO is an attractive approach when dealing with a protein for which the crystal structure has been determined and, therefore, HMFO is a suitable candidate. The aim of this work was to obtain a stable HMFO variant which performs well in the conversion of HMF into FDCA.
Results and discussion
The in silico phase of the FRESCO strategy was necessary to create an enriched library of single point mutants of HMFO. The crystal structure of the reduced form of the enzyme (PDB:4UDQ) was selected as model as it was solved at a better resolution (1.6 Å) than the oxidized enzyme . Using this structure, all possible point mutations were modeled (excluding residues that are within 5 Å from the FAD cofactor) and their respective values in free energy of folding were compared with that of the wild-type enzyme (∆∆GFold). By omitting the residues that are close to the active site, the risk of creating a mutant with lower or no activity is limited. The first in silico step consisted of a selection based on free energy prediction: single mutants with a predicted ∆∆GFold higher than − 5 kJ mol−1 were discarded. This decreased the number of variants to screen from 9044 to 744. The dynamic disulfide discovery (DDD) algorithm was not included in this FRESCO approach, because disulfide bonds may complicate protein expression . The MD simulations of the 744 variants were visually inspected comparing the modeled structure of the wild-type with the mutant ones. The goal was to select variants with putative improvements in their thermostability profile. The screening was based on avoiding features that are normally found to cause a decrease in stability such as an increased flexibility (backbone and sidechain), an increase in hydrophobic surface exposure and diminished hydrogen-bonding interactions . Those 140 mutants that scored well in the free energy of folding and did not show such aberrant structural features upon MD simulation and visual inspection were selected to be experimentally tested for thermostability.
Screening of single mutants
Golden Gate gene shuffling
The first step was the design of two synthetic genes, one wild-type version and one with the selected 7 mutations; each containing 8 BsaI restriction sites that flank the 7 gene modules that contain the target mutation sites. The positions of the modules were defined identically for the two versions and were designed, such that all the desired mutation sites could be introduced in separate modules and that the four-nucleotide overhangs for the ligation were unique and not palindromic. The innovation of this method is in the design of the BsaI restriction sites. Each module is flanked by two BsaI sites: between two modules there are two mirrored BsaI sites and the four overhang nucleotides at the end of each module are replicated at the beginning of the following module to allow a scarless ligation. By performing the one-pot Golden Gate cloning reaction on the mixture of the two donor vectors, containing the two synthetic genes, and the acceptor pBAD vector, all possible combinations of the 7 gene fragments are created in the pBAD-based expression vector. The Golden Gate gene shuffling library was analyzed by sequencing to confirm the heterogeneity and accuracy of the shuffling of gene fragments. The results showed that 97% of the colonies contained the correct restriction pattern and the shuffling efficiency was 65%. This indicates that per 100 clones, 65 contain non-redundant randomly shuffled sequences.
Shuffled library screening
Steady-state parameters of the different HMFO variants on HMF
kcat/Km (s−1 mM−1)
Engineering of thermostable HMFO variant for FDCA production
T m app values of HMFO variants
T m app (°C)
In this work, we further developed the FRESCO protocol by developing a novel and efficient approach for combining individual stabilizing mutations. The Golden Gate gene shuffling described in this study was fundamental to rapidly identify the best combination of thermostabilizing mutations that led to a stable and active HMFO variant. The engineered 7xHMFO variant presented a T m app improvement of 12 °C compared with the wild-type enzyme together with an improved cosolvent tolerance. We could also demonstrate that this thermostable variant of HMFO can be used as template to introduce a destabilizing mutation in the active site. One of the resulting HMFO mutants displayed superior performance compared to previously reported HMFO variants in converting HMF in a three-step one-enzyme reaction into FDCA.
All materials were acquired from Sigma-Aldrich unless otherwise specified.
The FRESCO method was employed to obtain a thermostabilized variant of HMFO. Computational modeling was performed using the 4UDQ X-ray structure of HMFO (1.6 Å resolution). To avoid mutations that could interfere with the active site, only residues that were > 5 Å away from FAD were mutated . The computational selection was started with calculating the predicted change of free folding energy (ΔΔGFold) with FoldX (foldx.crg.es) and Rosetta-ddg (http://www.rosettacommons.org) [18, 19, 31]. For Rosetta, the so-called row-3 protocol (described by Kellogg et al. in row 3 of their table 1) was invoked using the following options: -ddg::weight_file soft_rep_design -ddg::iterations 50 -ddg::local_opt_only true -ddg::min_cst false -ddg::mean true -ddg::min false -ddg::sc_min_only false -ddg::ramp_repulsive false -ddg::opt_radius 8.0 . For FoldX, the used options were --command=BuildModel --numberOfRuns=5. The single point mutations with a predicted ΔΔGFold < − 5 kJ mol−1 were subsequently submitted to MD simulations under Yasara as previously described [18, 19]. The averaged structures from the MD trajectories were visually inspected comparing the variant simulations with the wild-type HMFO while examining backbone and sidechain flexibility, hydrogen bonds, and hydrophobic exposure. This last in silico step is to further reduce the number of potentially thermostable variants to experimentally screen. All FRESCO specific scripts and code are available at https://www.rug.nl/staff/h.j.wijma.
For all the experiments, the His6x-SUMO-HMFO fusion has been used as it has been demonstrated before that the SUMO protein fused at the N-terminus does not affect the activity nor the thermostability of HMFO . The gene single point variants, the double mutant I73V H74Y, the 8AxHMFO, and the 8BxHMFO were obtained from pET SUMO-HMFO or pBAD SUMO-7xHMFO by whole-plasmid PCR with PfuUltra II Hotstart PCR Master Mix (Agilent). Template DNA was cleaved with DpnI (New England Biolabs) for at least 2 h at 37 °C. Escherichia coli NEB 10β (New England Biolabs) chemically competent cells were transformed (heat shock at 41 °C for 45 s) and cells plated on 50 μg mL−1 kanamycin or 100 μg mL−1 ampicillin LB agar plates. All mutations were confirmed by sequencing.
Golden Gate gene shuffling
The eightfold mutants were obtained using a gene shuffling approach. For this, we designed two synthetic gene versions of hmfo: one with 7 modules each containing one mutated region (the first module containing 2 mutations, and the 8 mutations are: I73V-H74Y, Q187E, G356H, V367L, T414K, A419Y, A435E) and the other synthetic gene with the same modules arrangement but without mutations, corresponding to the wild-type DNA sequence. Each module had been designed to be flanked by BsaI recognition sites: NGAGACC at the beginning and GGTCTCN at the end. Moreover, at the beginning of each fragment, the last 4 base pairs (bp) of the previous module are repeated (or the 5′-4 bp of the overhang region of the acceptor plasmid in the case of the first fragment and at the end of the 7th fragment are replicated 4 bp of the 3′-ligation site of the receiving plasmid). The BsaI cutting sites have been chosen to be unique (at least 3 out of 4 nucleotides in the sticky end have to be different) and not palindromic to avoid unwanted ligations (Additional file 9: Gene Sequences). The synthetic genes have been ordered (GenScript) cloned in two pUK57. A derivative of pBAD SUMO vector designed with BsaI cutting site 5′-TGGTngagacc and ggtctcnCTTG-3′) was used as receiving vector. The restriction–ligation reaction was set up in 20 μL volume with the components: pBAD SUMO 3.75 ng μL−1, pUC57wt 2.5 ng μL−1, pUC578xmutant 2.5 ng μL−1, T4 DNA ligase buffer (Promega), T4 DNA ligase 1.5 U μL−1 (Promega), BsaI 1 U μL−1 (New England Biolabs); the thermocycler program was: incubation at 37 °C for 5 min and at 16 °C for 10 min repeated 50 times, followed by a final incubation at 50 °C for 10 min (final digestion) and at 80 °C for 10 min (enzyme inactivation). The restriction–ligation reaction (5 μL) were used to transform 100 μL of chemically competent NEB 10β cells plated on 100 μg mL−1 ampicillin LB agar plates. The hmfo variants were verified first by colony PCR (DreamTaq Green PCR Master Mix, Thermo Fisher) to determine the inserts size and then by sequencing.
Large-scale expression and purification
For HMFO expression, a culture was started by inoculating 5 mL of preculture (LB supplemented with 50 μg mL−1 kanamycin or 100 μg mL−1 ampicillin) in 100 mL TB (Terrific Broth with the same antibiotic). In the case of pET constructs, protein expression was induced at OD600 0.5 with 1.0 mM isopropyl β-d-1-thiogalactopyranoside (IPTG), and with 0.02% l-arabinose in the case of the pBAD SUMO-HMFO construct. The latter was obtained by cloning the HMFO DNA sequence using NdeI and HindIII (NEB) in pBAD SUMO. Cells were grown overnight at 24 °C 135 rpm and harvested at 5000g for 10 min at 4 °C. The cells pellet was resuspended in 10 mL of 50 mM Tris HCl pH 8.0 with 150 mM NaCl and sonicated 3″ on 6″ off at 70% amplitude. The enzyme was purified from the cell-free extract as described previously .
The single point variants were expressed in E. coli BL21(DE3) cells with the pET SUMO-HMFO vector. The cultures were prepared using 600 μL of overnight culture (LB supplemented with 50 μg mL−1 kanamycin) to inoculate 5 mL TB containing 50 μg mL−1 kanamycin in a 24-well plate. The multi-site variants were expressed with the pBAD SUMO-HMFO vector in E. coli NEB 10β cells; the cultures were made by inoculating 200 μL of overnight culture (LB supplemented with 100 μg mL−1 ampicillin) in 800 μL of TB (100 μg mL−1 ampicillin) in 96-well plate. The cultures were incubated at 37 °C with a shaking at 300 rpm and induced with 1.0 mM IPTG [in case of E. coli BL21(DE3) cells carrying the pET SUMO-HMFO vector] or with 0.02% l-arabinose (E. coli NEB 10β cells carrying the pBAD SUMO-HMFO vector) at an optical density at 600 nm (OD600) of 2.0. Expression continued overnight at 24 °C with shaking at 550 rpm.
Cells were harvested by centrifugation at 2250g for 20 min at 4 °C. The cell-free extract was obtained after cell lysation: the cell pellet was solubilized in 200 μL lysis buffer (lysozyme 1 mg mL−1, deoxyribonuclease I 0.5 mg mL−1, MgCl2 10 mM in 50 mM Tris HCl pH 8.0). The solubilized pellet was incubated for 30 min at 25 °C (shaking at 550 rpm), and then, it was frozen in liquid nitrogen and centrifuged at 2250g for 45 min at 4 °C. The soluble fraction was filtered (Whatman UNIFILTER 96-well Microplate, GE-Healthcare) at 7g for 15 s at 4 °C and mixed with 100 μL of pre-equilibrated Ni-Sepharose resin (GE-Healthcare) for 15 min using an AcroPrep Advance 1 mL 96-well plate (Pall). The flow through was removed and the column was washed 2 times with 200 μL of 50 mM Tris HCl pH 8.0 with 150 mM NaCl, and one time with the same buffer containing 5 mM imidazole. The protein was eluted with 100 μL of 50 mm Tris HCl with 150 mM NaCl containing 500 mM imidazole. The eluate was desalted in 50 mM phosphate buffer pH 8.0 using PD MultiTrap G-25 plates (GE-Healthcare).
The melting temperature of HMFO cell-free extract or purified variants was tested by the ThermoFAD method, which allows to determine the unfolding temperature based on the release of the flavin cofactor [8, 24]. This assay was performed using 20 μL of CFE or 20 μL of purified enzyme in 50 mM phosphate buffer at pH 8.0. The ThermoFAD was also used to determine enzyme concentration after the small-scale purification based on the dT/fluorescence value using a calibration line. The calibration curve prepared with several WT concentrations proved that enzyme concentration does not affect the Tmapp.
All the activity assays were performed in 50 mM phosphate buffer pH 8.0 at 25 °C. The HMFO mutants I73V, H74Y, Q187E, G356H, V367L, T414K, A419Y, A435E were tested using enzyme activity towards vanillyl alcohol as reported previously . The gene shuffling library was tested with the coupled H2O2 detection assay: horseradish peroxidase (HRP) (Sigma), 0.004 U μL−1, 4-aminoantipyrine (0.1 mM), 3,5-dichloro-2-hydroxybenzenesulfonic acid (1 mM), and HMF (10 mM), measuring at 515 nm (ε515 = 26 mM−1 cm−1) the formation of pink product due to H2O2 production during the oxidation of HMF by HMFO. This HRP coupled assay was also used to determine kcat, Km, and the activity after the incubation at 40 °C for HMFO wild type and for the thermostable.
The conversion performed by HMFO WT, HMFO V367R-W466F, HMFO I73V-H74Y-G356H-V367L-T414K-A419Y-A435E (7xHMFO), HMFO I73V-H74Y-G356H-V367L-T414K-A419Y-A435E-W466F (8AxHMFO), HMFO I73V-H74Y-G356H-V367R-T414K-A419Y-A435E-W466F (8BxHMFO), using 5.0 mM 5-(hydroxymethyl)furfural as a substrate were carried out at 25 or 40 °C, with shaking at 1000 rpm. After the conversion, the enzyme was inactivated at 80 °C for 10 min and eliminated by centrifugation. The products were analyzed by high-performance liquid chromatography as described previously .
MWF and CM conceived the idea for this study. CM and HJW performed the computational modeling. CM designed the gene shuffling approach. CM and AOM optimized the library screening and performed the experiments. CM wrote the first draft of the manuscript. CM and MWF discussed the results and wrote the manuscript. All authors read and approved the final manuscript.
AOM received support from CONACYT (Consejo Nacional de Ciencia y Tecnología) and I2T2 (Instituto de Innovación y Transferencia de Tecnología-Nuevo León).
The authors declare that they have no competing interests.
Ethics approval and consent to participate
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
- Corma A, Iborra S, Velty A. Chemical routes for the transformation of biomass into chemicals. Chem Rev. 2007;107(6):2411–502.View ArticleGoogle Scholar
- Moreau C, Naceur M, Gandini A. Recent catalytic advances in the chemistry of substituted furans from carbohydrates and in the ensuing polymers. Top Catal. 2004;27:11–30.View ArticleGoogle Scholar
- Lima S, Antunes MM, Pillinger M, Valente AA. Ionic liquids as tools for the acid-catalyzed hydrolysis/dehydration of saccharides to furanic aldehydes. ChemCatChem. 2011;3:1686–706.View ArticleGoogle Scholar
- Zhang Z, Deng K. Recent advances in the catalytic synthesis of 2,5-furandicarboxylic acid and its derivatives. ACS Catal. 2015;5:6529–44.View ArticleGoogle Scholar
- Verdeguer P, Merat N, Gaset A. Oxydation catalytique du HMF en acide 2,5-furane dicarboxylique. J Mol Catal. 1993;85:327–44.View ArticleGoogle Scholar
- Van Putten R, Van Der Waal JC, De Jong E, Rasrendra CB, Heeres HJ, De Vries JG. Hydroxymethylfurfural, a versatile platform chemical made from renewable resources. Chem Rev. 2013;113(3):1499–597.View ArticleGoogle Scholar
- Yuan H, Li J, Shin H, Du G, Chen J, Shi Z, et al. Improved production of 2,5-furandicarboxylic acid by overexpression of 5-hydroxymethylfurfural oxidase and 5-hydroxymethylfurfural/furfural oxidoreductase in Raoultella ornithinolytica BF60. Bioresour Technol. 2018;247:1184–8.View ArticleGoogle Scholar
- Dijkman WP, Fraaije MW. Discovery and characterization of a 5-hydroxymethylfurfural oxidase from Methylovorus sp. strain MP688. Appl Environ Microbiol. 2014;80:1082–90.View ArticleGoogle Scholar
- Carro J, Ferreira P, Rodríguez L, Prieto A, Serrano A, Balcells B, et al. 5-hydroxymethylfurfural conversion by fungal aryl-alcohol oxidase and unspecific peroxygenase. FEBS J. 2015;282:3218–29.View ArticleGoogle Scholar
- Dijkman WP, Groothuis DE, Fraaije MW. Enzyme-catalyzed oxidation of 5-hydroxymethylfurfural to furan-2,5-dicarboxylic acid. Angew Chem Int Ed. 2014;53:6515–8.View ArticleGoogle Scholar
- Dijkman WP, Binda C, Fraaije MW, Mattevi A. Structure-based enzyme tailoring of 5-hydroxymethylfurfural oxidase. ACS Catal. 2015;5:1833–9.View ArticleGoogle Scholar
- Kristjansson JK. Thermophilic organisms as sources of thermostable enzymes. Trends Biotechnol. 1989;7:349–53.View ArticleGoogle Scholar
- Bloom JD, Labthavikul ST, Otey CR, Arnold FH. Protein stability promotes evolvability. Proc Natl Acad Sci. 2006;103:5869–74.View ArticleGoogle Scholar
- Bommarius AS, Broering JM, Chaparro-Riggers JF, Polizzi KM. High-throughput screening for enhanced protein stability. Curr Opin Biotechnol. 2006;17:606–10.View ArticleGoogle Scholar
- Turner NJ. Directed evolution drives the next generation of biocatalysts. Nat Chem Biol. 2009;5:567–73.View ArticleGoogle Scholar
- Eijsink VGH, Bjørk A, Gåseidnes S, Sirevåg R, Synstad B, Van Den Burg B, et al. Rational engineering of enzyme stability. J Biotechnol. 2004;113:105–20.View ArticleGoogle Scholar
- Steiner K, Schwab H. Recent advances in rational approaches for enzyme engineering. Comput Struct Biotechnol J. 2012. https://doi.org/10.5936/csbj.201209010.Google Scholar
- Wijma HJ, Floor RJ, Jekel PA, Baker D, Marrink SJ, Janssen DB. Computationally designed libraries for rapid enzyme stabilization. Protein Eng Des Sel. 2014;27:49–58.View ArticleGoogle Scholar
- Wijma HJ, Fürst MJLJ, Janssen DB. A computational library design protocol for rapid improvement of protein stability: FRESCO. In: Bornscheuer UT, Höhne M, editors. Protein engineering: methods and protocols. New York: Springer; 2018. p. 69–85.View ArticleGoogle Scholar
- Floor RJ, Wijma HJ, Colpa DI, Ramos-Silva A, Jekel PA, Szymański W, et al. Computational library design for increasing haloalkane dehalogenase stability. ChemBioChem. 2014;15:1660–72.View ArticleGoogle Scholar
- Wu B, Wijma HJ, Song L, Rozeboom HJ, Poloni C, Tian Y, et al. Versatile peptide C-terminal functionalization via a computationally engineered peptide amidase. ACS Catal. 2016;6:5405–14.View ArticleGoogle Scholar
- Arabnejad H, Lago MD, Jekel PA, Floor RJ, Thunnissen AWH, Van Scheltinga ACT, et al. A robust cosolvent-compatible halohydrin dehalogenase by computational library design. Protein Eng Des Sel. 2017;30:175–89.Google Scholar
- Nosoh Y, Sekiguchi T. Protein stability and stabilization through protein engineering. E. Horwood: Billingham; 1991.Google Scholar
- Forneris F, Orru R, Bonivento D, Chiarelli LR, Mattevi A. ThermoFAD, a Thermofluor-adapted flavin ad hoc detection system for protein folding and ligand binding. FEBS J. 2009;276:2833–40.View ArticleGoogle Scholar
- Ness JE, Kim S, Gottman A, Pak R, Krebber A, Borchert TV, et al. Synthetic shuffling expands functional protein diversity by allowing amino acids to recombine independently. Nat Biotechnol. 2002;20:1251–5.View ArticleGoogle Scholar
- Coco WM, Levinson WE, Crist MJ, Hektor HJ, Darzins A, Pienkos PT, et al. DNA shuffling method for generating highly recombined genes and evolved enzymes. Nat Biotechnol. 2001;19:354–9.View ArticleGoogle Scholar
- Hughes RA, Miklos AE, Ellington AD. Gene synthesis: methods and applications. Dublin: Academic Press; 2011.View ArticleGoogle Scholar
- Engler C, Kandzia R, Marillonnet S. A one pot, one step, precision cloning method with high throughput capability. PLoS ONE. 2008. https://doi.org/10.1371/journal.pone.0003647.Google Scholar
- Engler C, Gruetzner R, Kandzia R, Marillonnet S. Golden gate shuffling: a one-pot DNA shuffling method based on type IIs restriction enzymes. PLoS ONE. 2009. https://doi.org/10.1371/journal.pone.0005553.Google Scholar
- Zhang F, Cong L, Lodato S, Kosuri S, Church GM, Arlotta P. Efficient construction of sequence-specific TAL effectors for modulating mammalian transcription. Nat Biotechnol. 2011;29:149–54.View ArticleGoogle Scholar
- Guerois R, Nielsen JE, Serrano L. Predicting changes in the stability of proteins and protein complexes: a study of more than 1000 mutations. J Mol Biol. 2002;2836:369–87.View ArticleGoogle Scholar
- Kellogg EH, Leaver-Fay A, Baker D. Role of conformational sampling in computing mutation-induced changes in protein structure and stability. Proteins Struct Funct Bioinform. 2011;79:830–8.View ArticleGoogle Scholar